Language Understanding and Commnication, Affective Computing, Computer Science, Intelligent robot, Social Computing
Subject of Study
○
Emotion Robort, Machine Aided Writing, Natural Language Processing and Knowledge Engineering, Sentience Computer and Sensitivity Technique, Multi-Lingual Multi-Function Multi-Media Intelligent System Project, Machine Translation and Information Retrieval (natural language processing, knowledge engineering, sentience computer, machine translation, Machine-Aided English Writing, automatic abstracting, Dialogue machine translation, information retrieval)
Book / Paper
Book:
1.
Fuji Ren and Kazuyuki Matsumoto : 言語・音声・顔表情・脳波を総合利用した感情測定システム, CMC Publishing Co.,Ltd., Aug. 2019.
(Keyword)
multi modal sensing / emotion measurement system / voice / language
2.
Xiaohua Wang, Muzi Peng, Lijuan Pan, Min Hu, Chunhua Jin and Fuji Ren : Two-Level Attention with Multi-task Learning for Facial Emotion Estimation, Springer-Verlag, Dec. 2018.
(Summary)
Valence-Arousal model can represent complex human emotions, including slight changes of emotion. Most prior works of facial emotion estimation only considered laboratory data and used video, speech or other multi-modal features. The effect of these methods applied on static images in the real world is unknown. In this paper, a two-level attention with multi-task learning (MTL) framework is proposed for facial emotion estimation on static images. The features of corresponding region were automatically extracted and enhanced by first-level attention mechanism. And then we designed a practical structure to process the features extracted by first-level attention. In the following, we utilized Bi-directional Recurrent Neural Network (Bi-RNN) with self-attention (second-level attention) to make full use of the relationship of these features adaptively. It can be concluded as a combination of global and local information. In addition, we exploited MTL to estimate the value of valence and arousal simultaneously, which employed the correlation of the two tasks. The quantitative results conducted on AffectNet dataset demonstrated the superiority of the proposed framework. In addition, extensive experiments were carried out to analysis effectiveness of different components.
Yu Gu, Tao Liu, Jie Li, Fuji Ren, Zhi Liu, Xiaoyan Wang and Peng Li : EmoSense: Data-driven Emotion Sensing via Off-the-shelf WiFi Devices, IEEE, Kansas City, USA, May 2018.
Yu Gu, Min Peng, Fuji Ren and Jie Li : Smart Technologies for Emergency Response and Disaster Management (Chaoter 3, WiFi Fingerprint Localization for Emergency Response: Harvesting Environmental Dynamics for a Rapid Setup), IGI Global, Jan. 2018.
Fuji Ren, Shun Nishide, Kazuyuki Matsumoto, XIN KANG, Duo Feng and Mengjia He : Artificial Intelligence with Uncertainty, NTS, Jun. 2017.
6.
Fuji Ren and Kazuyuki Matsumoto : Natural Language Processing Capabilities Required for Humanoid Nursing Robots, Fukuro Shuppan Publishing, Mar. 2017.
7.
Kazuyuki Matsumoto, Fuji Ren, Minoru Yoshida and Kenji Kita : Refinement by Filtering Translation Candidates and Similarity Based Approach to Expand Emotion Tagged Corpus, Jan. 2017.
(Summary)
We attempted to expand corpus without translating target linguistic resource. The result of the evaluation experiment using the machine learning algorithm showed the effectiveness of the expanded emotion corpus based on the original languages unannotated sentences and their similar sentences.
(Keyword)
emotion tagged corpus / Japanese-English parallel corpora / emotion estimation
Zhongzhi Shi, Ben Goertzel and Fuji Ren : Advanced Intelligence, Tsinghua University Press,, Beijing, Aug. 2010.
9.
Fuji Ren : Language Engineering, Affective Computing and Advanced Intelligence, BUPT publishing House, Dec. 2009.
10.
Ye Wu and Fuji Ren : A Corpus-based Multi-label Emotion Classification using Maximum Entropy, --- 20. Natural Language Processing and Cognitive Science, B. Sharp. and M. Zock (Eds.) ---, INSTICC Press, Jun. 2009.
11.
Fuji Ren : Advanced Intelligence, International Advanced Information Institute, Dec. 2008.
12.
Fuji Ren, Yixin Zhong, Shingo Kuroiwa and Satoru Tsuge : Artificial Intelligence and Affective Computing, International Advanced Information Institute, Jun. 2007.
Fan Lixin, Fuji Ren, Miyanaga Yoshikazu and Tochinai Koji : Automatic Composition of Chinese Compound Words for Chinese-Japanese Machine Translation System, Computational Linguistics, Nov. 1993.
Fuji Ren, Yangyang Zhou, Jiawen Deng, Kazuyuki Matsumoto, Duo Feng, Tianhao She, Ziyun Jiao, Zheng Liu, Taihao Li, Satoshi Nakagawa and XIN KANG : Tracking Emotions using an Evolutionary Model of Mental State Transitions: Introducing a New Paradigm, Intelligent Computing, 1-24, 2024.
LIU ZHENG, XIN KANG and Fuji Ren : Dual-TBNet: Improving the Robustness of Speech Features via Dual-Transformer-BiLSTM for Speech Emotion Recognition, IEEE/ACM Transactions on Audio, Speech, and Language Processing, Vol.31, 2193-2203, 2023.
Fei Ding, XIN KANG and Fuji Ren : Neuro or Symbolic? Fine-tuned Transformer with Unsupervised LDA Topic Clustering for Text Sentiment Analysis, IEEE Transactions on Affective Computing, 1-15, 2023.
XIN KANG, Shi Xuefeng, Yunong Wu and Fuji Ren : Active learning with complementary sampling for instructing class-biased multi-label text emotion classification, IEEE Transactions on Affective Computing, Vol.14, No.1, 523-536, 2023.
(Keyword)
Active Learning / Complementary Sampling / Class Biased Multi Label Classification / Text Emotion
Fuji Ren, Zheng Liu and XIN KANG : An efficient framework for constructing speech emotion corpus based on integrated active learning strategies, IEEE Transactions on Affective Computing, Vol.13, No.4, 1929-1940, 2022.
Li Zhiyang, Li Bin, Ni Hongjun, Fuji Ren, Lv Shuaishuai and XIN KANG : An Effective Surface Defect Classification Method Based on RepVGG with CBAM Attention Mechanism (RepVGG-CBAM) for Aluminum Profiles, Metals, Vol.12, No.11, 1-16, 2022.
Kazuyuki Matsumoto, Manabu Sasayama, Minoru Yoshida, Kenji Kita and Fuji Ren : Emotion Analysis and Dialogue Breakdown Detection in Dialogue of Chat Systems Based on Deep Neural Networks, Electronics, Vol.11, No.5, 2022.
(Summary)
In dialogues between robots or computers and humans, dialogue breakdown analysis is an important tool for achieving better chat dialogues. Conventional dialogue breakdown detection methods focus on semantic variance. Although these methods can detect dialogue breakdowns based on semantic gaps, they cannot always detect emotional breakdowns in dialogues. In chat dialogue systems, emotions are sometimes included in the utterances of the system when responding to the speaker. In this study, we detect emotions from utterances, analyze emotional changes, and use them as the dialogue breakdown feature. The proposed method estimates emotions by utterance unit and generates features by calculating the similarity of the emotions of the utterance and the emotions that have appeared in prior utterances. We employ deep neural networks using sentence distributed representation vectors as the feature. In an evaluation of experimental results, the proposed method achieved a higher dialogue breakdown detection rate when compared to the method using a sentence distributed representation vectors.
(Keyword)
natural language processing / dialogue breakdown / human-computer dialogue system / sentiment analysis / emotion recognition
Qian Zhang and Fuji Ren : Double bayesian pairwise learning for one-class collaborative filtering, Knowledge-based Systems, Vol.440, 2-13, 2021.
(Summary)
Recommender systems have become an indispensable tool for real-world applications. Only one-class feedback can be obtained in many applications. Therefore, the one-class recommendation problem has attracted much attention. Pairwise ranking methods are popular for dealing with the one-class problem. Bayesian Personalized Ranking (BPR) is one of the most popular pairwise methods, assuming users prefer the observed item to the unobserved item. The parameters in BPR are learned based on stochastic gradient descent (SGD). However, the previous work has shown that existing the vanishing gradient problem in the learning process when the preference difference between the observed item and the unobserved item is very large. In this paper, we propose a novel algorithm called Double Bayesian Pairwise Learning (DBPL). In the learning process of DBPL, the preference difference between the observed item and the unobserved item can be reduced by fusing a relatively smaller preference difference between another pair of items. Moreover, we calculate potential preference scores between users and items based on user-item interactions to measure preference differences between unobserved items of each user. Experimental results on three real-world datasets show the effectiveness of the DBPL algorithm.
Siyuan Xue and Fuji Ren : Intent-Enhanced Attentive Bert Capsule Network for Zero-shot Intention Detection, Neurocomputing, 2-15, 2021.
(Summary)
Spoken language understanding (SLU) plays an indispensable role in the dialogue system. The traditional intention detection task is regarded as a classification problem where utterances are associated with pre-defined intents. However, the various expressions of user's intents and constantly emerging novel intents make the annotating time-consuming and labor-intensive, building massive obstacles for extending the model to new tasks. Identifying unexpected user intention and achieving the user's desire goal is a challenging task. Therefore, we conduct zero-shot intention detection based on a transformation-based learning manner. In this paper, we propose an intent-enhanced attentive capsule network (IE-BertCapsNet) further guides the aggregation process of the capsule network and generalizable useful features that can be adapted to emerging intentions. Coupling with the large margin cosine loss function, the proposed model can identify discriminative features by forcing the whole network to minimize inter-class distance and minimize intra-class distance. Finally, we leverage the IE-BertCapsNet's feature extraction ability and knowledge transferring capability to conduct zero-shot intent detection and generalized zero-shot intent detection. Extensive experiments on five benchmark task-oriented datasets in four languages demonstrate that the proposed model can achieve competitive performance that can better discriminate known intents and detect unknown intents.
Min Hu, Qian Chu, Xiaohua Wang, Lei He and Fuji Ren : A Two-Stage Spatiotemporal Attention Convolution Network for Continuous Dimensional Emotion Recognition from Facial Video, IEEE Signal Processing Letters, 1-20, 2021.
(Summary)
Continuous dimensional emotion recognition for facial video sequence is a crucial and challenging task in Affective Computing and Human-Computer Intelligent Interaction. The key of this task is to effectively extract and discriminate spatialtemporal features in a more fine-grained way. In this paper, a Two-Stage Spatiotemporal Attention Temporal Convolution Network (TS-SATCN) is designed for continuous dimensional emotion recognition of facial videos. The first stage generates an initial recognition result that is later fed into the second for correction. In each stage, the introduced spatiotemporal attention branch helps the network learn different attention levels and focuses on the informative spatial-temporal features adaptively. The network is trained by a proposed smooth loss function which can further improve the predictions quality. Extensive experiments are performed on two datasets, RECOLA and AFEW-VA, which shows that the proposed method achieves significant improvement over state-of-the-art methods.
Yu Gu, Huan Yan, Mianxiong Dong, Meng Wang, Xiang Zhang, Zhi Liu and Fuji Ren : WiONE: One-Shot Learning for Environment-Robust Device-Free User Authentication via Commodity Wi-Fi in Man-Machine System, IEEE Transactions on Computational Social Systems, 1-13, 2021.
(Summary)
User authentication is the first and most critical step in protecting a man-machine system from a malicious spoofer. However, security and privacy are just like the two sides of one coin, hard to see both at the same time, especially by the current mainstream credential- and biometric-based approaches. To this end, we propose WiONE, a safe and privacy-preserving user authentication system leveraging the ubiquitous Wi-Fi infrastructure by exploring ``how you behave'' rather than ``who you are''. The key idea is to apply deep learning to user physical behavior captured by Wi-Fi channel state information (CSI) to identify legitimate users while rejecting spoofers. The design of WiONE faces two challenges, namely, how to capture the subtle behavior, such as a keystroke on CSI, and how to mitigate the heavy environment-specific training required by deep learning. For the former, we design a behavior enhancement model based on the Rician fading to highlight the behavior-induced information by suppressing the behavior-unrelated information on channel response. For the latter, we develop a behavior characterization method tailored for the prototypical networks to facilitate the extraction of the domain-independent behavioral features and enable one-shot recognition of a new user in a new environment. Numerous experiments are conducted in several real-world environments, and the results show that WiONE outperforms its state-of-the-art rivals in authentication performance with much less training effort.
Min Hu, Fei Qian, Xiaohua Wang, Lei He, Dong Guo and Fuji Ren : Robust Heart Rate Estimation with Spatial-Temporal Attention Network from Facial Videos, IEEE Transactions on Cognitive and Developmental Systems, 1-20, 2021.
(Summary)
In order to solve the problems of highly redundant spatial information and motion noise in the heart rate (HR) estimation from facial videos based on remote Photoplethysmography(rPPG), this paper proposes a novel HR estimation method based on spatial-temporal attention model. Firstly, to reduce the redundant information and strengthen the association relationships of long-range videos, the spatial-temporal facial features are extracted by the 2D convolutional neural network (2DCNN) and 3D convolutional neural network (3DCNN), respectively. The aggregation function is adopted to incorporate feature maps into short segment spatial-temporal feature maps. Secondly, the spatial-temporal strip pooling is designed in the spatial-temporal attention module to reduce head movement noises. Then, via the two-part loss function, the model can focus more on the rPPG signal rather than the interference. We conduct extensive experiments on two public datasets to verify the effectiveness of our model. The experimental results show that the proposed method achieves significantly better performances than the state-of-the-art baselines: The mean absolute error could be reduced by 11% on the PURE dataset, and by 25% on the COHFACE dataset.
Qian Zhang and Fuji Ren : Prior-based bayesian pairwise ranking for one-class collaborative filtering, Neurocomputing, Vol.440, 365-374, 2021.
(Summary)
In many real-world applications, only user-item interactions (one-class feedback) can be observed. The recommendation methods have been studied for personalized ranking with one-class feedback in recent years. Pairwise ranking methods have been widely used for dealing with the one-class problem with the assumption that users prefer their observed items over unobserved items. However, existing some items that users have not seen yet. It is unsuitable for treating all unobserved items of the user as negative feedback. In this paper, we propose a Prior-based Bayesian Pairwise Ranking (PBPR) model, which relaxes the simple pairwise preference assumption in previous works by further considering the pairwise preference between two unobserved items. Moreover, we calculate users' potential preference scores on unobserved items, i.e., prior information, based on historical interactions. The prior information can be used to measure the fine-grained preference difference between any two unobserved items of each user. Through extensive experiments on real-world datasets, we demonstrate the effectiveness of our proposed recommendation method.
Jiawen Deng and Fuji Ren : A Survey of Textual Emotion Recognition and Its Challenges, IEEE Transactions on Affective Computing, 1-20, 2021.
(Summary)
Textual language is the most natural carrier of human emotion. In natural language processing, textual emotion recognition (TER) has become an important topic due to its significant academic and commercial potential. With the advanced development of deep learning technologies, TER has attracted growing attention and has significantly been promoted in recent years. This paper provides a systematic survey of latest TER advances, focusing on approaches using deep neural networks. According to how deep learning works at each stage, TER approaches are reviewed on word embedding, architecture, and training levels, respectively. We discussed the remaining challenges and opportunities from four aspects: the shortage of large-scale and high-quality dataset, fuzzy emotional boundaries, incomplete extractable emotional information in texts, and TER in dialogue. This paper creates a systematic and in-depth overview of deep TER technologies. It provides the necessary knowledge and new insights for relevant researchers to understand better the research state, remaining challenges, and future directions in this field.
Haitao Yu, Degen Huang, Fuji Ren and Lishuang Li : Diagnostic Evaluation of Policy-Gradient-Based Ranking, Electronics, Vol.10, No.19, 1-21, 2021.
(Summary)
Learning-to-rank has been intensively studied and has shown significantly increasing values in a wide range of domains, such as web search, recommender systems, dialogue systems, machine translation, and even computational biology, to name a few. In light of recent advances in neural networks, there has been a strong and continuing interest in exploring how to deploy popular techniques, such as reinforcement learning and adversarial learning, to solve ranking problems. However, armed with the aforesaid popular techniques, most studies tend to show how effective a new method is. A comprehensive comparison between techniques and an in-depth analysis of their deficiencies are somehow overlooked. This paper is motivated by the observation that recent ranking methods based on either reinforcement learning or adversarial learning boil down to policy-gradient-based optimization. Based on the widely used benchmark collections with complete information (where relevance labels are known for all items), such as MSLRWEB30K and Yahoo-Set1, we thoroughly investigate the extent to which policy-gradient-based ranking methods are effective. On one hand, we analytically identify the pitfalls of policy-gradient-based ranking. On the other hand, we experimentally compare a wide range of representative methods. The experimental results echo our analysis and show that policy-gradient-based ranking methods are, by a large margin, inferior to many conventional ranking methods. Regardless of whether we use reinforcement learning or adversarial learning, the failures are largely attributable to the gradient estimation based on sampled rankings, which significantly diverge from ideal rankings. In particular, the larger the number of documents per query and the more fine-grained the ground-truth labels, the greater the impact policy-gradient-based ranking suffers. Careful examination of this weakness is highly recommended for developing enhanced methods based on policy gradient.
Min Hu, Dong Guo, Mingxing Jiang, Fei Qian, Xiaohua Wang and Fuji Ren : rPPG-Based Heart Rate Estimation using Spatial-Temporal Attention Network, IEEE Transactions on Cognitive and Developmental Systems, 1-20, 2021.
(Summary)
Remote photoplethysmography (rPPG) based on computer vision technology is widely used to calculate the heart rate (HR) from facial videos. Existing rPPG techniques have been subject to some limitations (e.g., highly redundant spatial information, head movement noise and region of interest (ROI) selection). To address these limitations, this paper introduces an effective spatial-temporal attention network. A temporal fusion module is firstly proposed to fully exploit the time-domain information, aiming to reduce the redundant video information and strengthen the association relationships of long-range videos. Furthermore, a spatial attention mechanism is designed in the backbone net to precisely target the skin ROIs. Finally, to assist the network in learning the weights between channels, we project the RGB images using plane orthogonal to skin (POS) algorithm and add motion representation to complement physiological signals' extraction. Extensive experiments on the public PURE, MMSE-HR, and UBFC-rPPG datasets demonstrate that our model achieve competitive results compared with other methods.
Mingxing Jiang, Liquan Shen, Min Hu, Ping An and Fuji Ren : Blind Quality Evaluator of Tone-Mapped HDR and Multi-Exposure Fused Images for Electronic Display, IEEE Transactions on Consumer Electronics, 1-13, 2021.
(Summary)
The problem of reproducing high dynamic range (HDR) images on electronic display and photography with restricted dynamic range has gained a lot of interest in the consumer electronics community. There exist various approaches to this issue, e.g., tone mapping operators (TMOs) and multi-exposure fusion algorithms (MEFs). Many existing image quality assessment (IQA) methods have been proposed to compare images of quality degradation generated by TMOs/MEFs. Although promising performances have been achieved, they seldom consider local specific artifacts difference (i.e., abnormal exposure and color cast) related with the TMOs/MEFs. To address this limitation, this paper proposes a Blind Quality Evaluator of Tone-Mapped HDR and Multi-Exposure Fused Images (BQE-TM/MEFI). First, two purpose-designed segment models are utilized to distinguish well-exposedness dense patches (WEDPes) and non-WEDPes, color cast patches (CCPes) and non-CCPes respectively. Second, multiple quality-perception features are extracted to measure local artifacts: 1) structure and sharpness features from WEDPes, 2) saturation features from non-CCPes, and 3) edge structure features. Then, three new low-complexity regional features (over-exposure ratio, entropy and color confidence index) are calculated based on over-exposure segmentation model. Finally, all extracted features are aggregated into a machine-learning regression model to pool a quality score. The simplicity and good performance of the proposed method makes it suitable for electronic displays and other consumer electronics.
Yu Gu, Huan Yan, Xiang Zhang, Zhi Liu and Fuji Ren : 3D Facial Expression Recognition via Attention-based Multi-channel Data Fusion Network, IEEE Transactions on Instrumentation and Measurement, 1-11, 2021.
(Summary)
Facial expression has long been recognized as containing meaningful non-verbal affective cues for decoding human emotions. Recently, multi-modal 2D+3D fusion method has shown significant potential in facial expression recognition (FER) due to its fine-grained face descriptions in various spatial channels. However, current work mainly relies on feature-level or even score-level fusion to find emotion cues spread in different channels, and may miss key information due to lack of focus. To this end, we propose an attention-based multi-channel data fusion network (AMDFN) to better preserve and find such key facial cues. More specifically, we first map a 3D face scan into multi-channel images, and then fuse them in a ResNet18 backbone to get layered emotion features. Secondly, we leverage a layer attention model to explore the dependencies between features of different layers to learn discriminative affective cues for effective emotion recognition. Our comprehensive experiments on two widely-used datasets (i.e., Facescape and Bosphorus) have verified the performance of our approach compared to several state-of-the-art rivals.
Fuji Ren and Tianhao She : Utilizing External Knowledge to Enhance Semantics in Emotion Detection in Conversation, IEEE Access, Vol.10, No.19, 1-11, 2021.
(Summary)
Enabling machines to emotion recognition in conversation is challenging, mainly because information in human dialogue innately conveys emotions by long-term experience, abundant knowledge, context, and the intricate patterns between the affective states. We address the task of emotion recognition in conversations using external knowledge to enhance semantics. We propose KES model, a new framework that incorporates different elements of external knowledge and conversational semantic role labeling, where build upon them to learn interactions between interlocutors participating in a conversation. We design a self-attention layer specialized for enhanced semantic text features with external commonsense knowledge. Then, two different networks composed of LSTM are responsible for tracking individual internal state and context external state. In addition, the proposed model has experimented on three datasets in emotion detection in conversation. The experimental results show that our model outperforms the state-of-the-art approaches on most of the tested datasets.
Quan Wang, Fei Wang, Fuji Ren, Zhongheng Li and Feiping Nie : An Effective Clustering Optimization Method for Unsupervised Linear Discriminant Analysis, IEEE Transactions on Knowledge and Data Engineering, 1-13, 2021.
(Summary)
The recent work Unsupervised Linear Discriminant Analysis (Un-LDA) completes its clustering process during the alternating optimization by converting equivalently the objective and finally using the K-means algorithm. However, the K-means algorithm has its inherent drawbacks. It is hard for the K-means algorithm to deal well with some complex clustering cases where there are too many real clusters or non-convex clusters. In this paper, a novel clustering optimization method is presented to accomplish the clustering process in Un-LDA and the resulting method can be named Un-LDA(CD). Specifically, instead of the K-means algorithm, an elaborately designed coordinate descent algorithm is adopted to obtain the clusters after the objective function goes through a series of simple but deft equivalent conversions. Extensive experiments have demonstrated that the coordinate descent clustering solution for Un-LDA can outperform the original K-means based solution on the tested data sets especially those complex data sets with a pretty large number of real clusters.
Tianhao She and Fuji Ren : Enhance the Language Ability of Humanoid Robot NAO through Deep Learning to Interact with Autistic Children, Electronics, Vol.10, No.19, 1-21, 2021.
(Summary)
Autism spectrum disorder (ASD) is a life-long neurological disability, and a cure has not yet been found. ASD begins early in childhood and lasts throughout a person's life. Through early intervention, many actions can be taken to improve the quality of life of children. Robots are one of the best choices for accompanying children with autism. However, for most robots, the dialogue system uses traditional techniques to produce responses. Robots cannot produce meaningful answers when the conversations have not been recorded in a database. The main contribution of our work is the incorporation of a conversation model into an actual robot system for supporting children with autism. We present the use a neural network model as the generative conversational agent, which aimed at generating meaningful and coherent dialogue responses given the dialogue history. The proposed model shares an embedding layer between the encoding and decoding processes through adoption. The model is different from the canonical Seq2Seq model in which the encoder output is used only to set-up the initial state of the decoder to avoid favoring short and unconditional responses with high prior probability. In order to improve the sensitivity to context, we changed the input method of the model to better adapt to the utterances of children with autism. We adopted transfer learning to make the proposed model learn the characteristics of dialogue with autistic children and to solve the problem of the insufficient corpus of dialogue. Experiments showed that the proposed method was superior to the canonical Seq2sSeq model and the GAN-based dialogue model in both automatic evaluation indicators and human evaluation, including pushing the BLEU precision to 0.23, the greedy matching score to 0.69, the embedding average score to 0.82, the vector extrema score to 0.55, the skip-thought score to 0.65, the KL divergence score to 5.73, and the EMD score to 12.21.
Wenjie Liu, Guoqing Wu and Fuji Ren : Deep Multibranch Fusion Residual Network for Insect Pest Recognition, IEEE Transactions on Cognitive and Developmental Systems, Vol.13, No.3, 705-716, 2021.
(Summary)
Earlier insect pest recognition is one of the critical factors for agricultural yield. Thus, an effective method to recognize the category of insect pests has become significant issues in the agricultural field. In this article, we proposed a new residual block to learn multiscale representation. In each block, it contains three branches: one is parameter-free, and the others contain several successive convolution layers. Moreover, we proposed a module and embedded it into the new residual block to recalibrate the channelwise feature response and to model the relationship of the three branches. By stacking this kind of block, we constructed the deep multibranch fusion residual network (DMF-ResNet). For evaluating the model performance, we first test our model on CIFAR-10 and CIFAR-100 benchmark data sets. The experimental results show that DMF-ResNet outperforms the baseline models significantly. Then, we construct DMF-ResNet with different depths for high-resolution image classification tasks and apply it to recognize insect pests. We evaluate the model performance on the IP102 data set, and the experimental results show that DMF-ResNet could achieve the best accuracy performance than the baseline models and other stateof- the-art methods. Based on these empirical experiments, we demonstrate the effectiveness of our approach.
Tian Chen, Hongfang Yin, Xiaohui Yuan, Yu Gu, Fuji Ren and Xiao Sun : Emotion recognition based on fusion of long short-term memory networks and SVMs, Digital Signal Processing, 1-10, 2021.
(Summary)
This paper proposes a multimodal fusion emotion recognition method based on Dempster-Shafer evidence theory, which includes electroencephalogram (EEG) and electrocardiogram (ECG). For EEG, we use the SVM classifier to classify features, and for ECG, we establish the corresponding Bi-directional Long Short-Term Memory network emotion recognition structure, which is fused with EEG classification results through the evidence theory. We selected 25 video clips with five emotions (happy, relaxed, angry, sad, and disgusted), and a total of 20 subjects participated in our emotional experiment. The experimental results prove that the performance of the multi-modal fusion model proposed in this paper is superior to the single-modal emotion recognition model. In the Arousal and Valance dimensions, the average accuracy is improved by 2.64% and 2.75% compared with the EEG signal-based emotion recognition model. Compared with the emotion recognition model based on the ECG signal, the accuracy is improved by 7.37% and 8.73%.
Duo Feng and Fuji Ren : Data-Driven Channel Pruning towards Local Binary Convolution Inverse Bottleneck Network Based on Squeeze-and-Excitation Optimization Weights, Electronics, 1-14, 2021.
(Summary)
This paper proposed a model pruning method based on local binary convolution (LBC) and squeeze-and-excitation (SE) optimization weights. We first proposed an efficient deep separation convolution model based on the LBC kernel. By expanding the number of LBC kernels in the model, we have trained a larger model with better results, but more parameters and slower calculation speed. Then, we extract the SE optimization weight value of each SE module according to the data samples and score the LBC kernel accordingly. Based on the score of each LBC kernel corresponding to the convolution channel, we performed channel-based model pruning, which greatly reduced the number of model parameters and accelerated the calculation speed. The model pruning method proposed in this paper is verified in the image classification database. Experiments show that, in the model using the LBC kernel, as the number of LBC kernels increases, the recognition accuracy will increase. At the same time, the experiment also proved that the recognition accuracy is maintained at a similar level in the small parameter model after channel-based model pruning by the SE optimization weight value.
Min Hu, Peng Ge, Xiaohua Wang, Hui Lin and Fuji Ren : A spatio-temporal integrated model based on local and global features for video expression recognition, The Visual Computer, 1-18, 2021.
Xiaohua Wang, Cong Yu, Yu Gu, Min Hu and Fuji Ren : Multi-Task and Attention Collaborative Network for Facial Emotion Recognition, IEEJ Transactions on Electrical and Electronic Engineering (TEEE), Vol.16, No.4, 568-576, 2021.
(Summary)
Facial expression is one of the most direct and effective ways to recognize emotions, widely used in human-computer interaction, affective computing, and other research fields. Expression recognition can be divided into discrete expression classification and continuous dimensional emotion recognition. Most of the existing multi-dimensional emotional estimation only considers the data under laboratory conditions. In this paper, facial emotion estimation is performed based on real-world images and combined with the advantages of multi-task learning and attention mechanism. We improve the multi-task attention network (MTAN) from two aspects: task and feature. At the aspect of the task, the multi-task collaborative attention network (MTCAN), which is based on task correlation, is proposed to solve task deviation in multi-task learning. At the aspect of the feature, based on MTCAN, we came up with MTACN, which used the self-attention mechanism to measure the importance of each attention module for each specific task. Then, we can capture the local-to-global connection in one step and fully exploit the feature within different levels of each task. Experimental results on the AffectNet dataset show that the performance of the model is significantly better than the original network, and the Root-mean-square error and consistency correlation coefficient results are superior to other existing models.
Yokotani Tomoya, Tanioka Ryuichi, Kawai Chihiro, Betriana Feni, Hirokazu Ito, Yuko Yasuhara, Tetsuya Tanioka, Rozzano De Castro Locsin, Kazuyuki Matsumoto and Fuji Ren : Human Psychological Burden and Thinking Process while Operating a Humanoid Robot Conversation App Program, International Journal of Advanced Intelligence (IJAI), Vol.12, No.1, 23-35, 2021.
(Keyword)
Older adults / communication robot / Pepper robot / natural language processing / artificial intelligence / human thinking process / psychological burden
33.
Hirokazu Ito, Kazuyuki Matsumoto, XIN KANG, Tetsuya Tanioka, Yuko Yasuhara, Rozzano De Castro Locsin and Fuji Ren : The AI Development Through Transforming Tacit Knowledge to Explicit Knowledge of Nurses' Dialogue for Patients with Dementia, International Journal of Advanced Intelligence (IJAI), Vol.12, No.1, 11-21, 2021.
(Keyword)
natural language processing / artificial intelligence / dementia care / PsyNACS / explicit and tacit knowledge / nurses' dialogue / humanoid robot
Yuki Obayashi, Feni Betriana, Tetsuya Tanioka, Tomoya Yokotani, Ryuichi Tanioka, Chihiro Kawai, Hirokazu Ito, Yuko Yasuhara, Rozzano De Castro Locsin, Kyoko Osaka, Kazuyuki Matsumoto, Fuji Ren and Yoshihiro Kai : Developmental Issues of Communication for Robot Applications in Older People Care: An Integrative Review, International Journal of Advanced Intelligence (IJAI), Vol.12, No.1, 53-67, 2021.
(Keyword)
Communication robots / applications / performance of humanoid robots / developmental issues
35.
Jun Liu, Tingting Wang and Fuji Ren : Lock Isolation Security Architecture for Reconfigurable Scanning Networks, Journal of Computer-Aided Design & Computer Graphics, Vol.33, No.3, 1-7, 2021.
(Summary)
To protect the reconfigurable scanning network from the influence of unauthorized access, tampering of transmitted data by malicious instruments and sniffing, a lock isolation security structure is proposed. Firstly, the presented structure divides the instruments without security threat to each other into one group, and the groups can be accessed one by one via the isolated signals, which can prevent malicious instruments tampering and sniffing transmitted data. Secondly, the key instruments are protected by the lock segment insertion bit. The lock segment insertion bit can only be opened when multiple key values at specific positions are set to specific values (0, 1 sequence), which can increase the difficulty of unauthorized access. Besides, to reduce the hardware overhead, and the difficulty of place and route caused by excessive instrument groups, instrument grouping algorithm is proposed. The proposed algorithm firstly constructs an undirected graph based on the security relationship among instruments. Then the maximal independent sets of the undirected are obtained. In this way, the number of instrument groups is reduced greatly. Compared to other similar methodologies, the experimental results conducted on ITC 02 benchmark circuits show that the time consumption for opening the locking segment insertion bit of the proposed security structure increased by 7 times, and the average percentage reduction of area, power consumption and wire length is 3.81%,
Min Hu, Fei Qian, Dong Guo, Xiaohua Wang, Lei He and Fuji Ren : ETA-rPPGNet: Effective Time-Domain Attention Network for Remote Heart Rate Measurement, IEEE Transactions on Instrumentation and Measurement, Vol.70, 1-12, 2021.
Jiawen Deng and Fuji Ren : Hierarchical Network with Label Embedding for Contextual Emotion Recognition, Research : A Science Partner Journal, Vol.2021, No.1, 1-9, 2021.
(Summary)
Emotion recognition has been used widely in various applications such as mental health monitoring and emotional management. Usually, emotion recognition is regarded as a text classification task. Emotion recognition is a more complex problem, and the relations of emotions expressed in a text are nonnegligible. In this paper, a hierarchical model with label embedding is proposed for contextual emotion recognition. Especially, a hierarchical model is utilized to learn the emotional representation of a given sentence based on its contextual information. To give emotion correlation-based recognition, a label embedding matrix is trained by joint learning, which contributes to the final prediction. Comparison experiments are conducted on Chinese emotional corpus RenCECps, and the experimental results indicate that our approach has a satisfying performance in textual emotion recognition task.
Ziyun Jiao and Fuji Ren : WRGAN: Improvement of RelGAN with Wasserstein Loss for Text Generation, Electronics, 1-14, 2021.
(Summary)
Generative adversarial networks (GANs) were rst proposed in 2014, and have been widely used in computer vision, such as for image generation and other tasks. However, the GANs used for text generation have made slow progress. One of the reasons is that the discriminator's guidance for the generator is too weak, which means that the generator can only get a ``true or false'' probability in return. Compared with the current loss function, the Wasserstein distance can provide more information to the generator, but RelGAN does not work well with Wasserstein distance in experiments. In this paper, we propose an improved neural network based on RelGAN and Wasserstein loss named WRGAN. Differently from RelGAN, we modi ed the discriminator network structure with 1D convolution of multiple different kernel sizes. Correspondingly, we also changed the loss function of the network with a gradient penalty Wasserstein loss. Our experiments on multiplepublicdatasetsshowthatWRGANoutperformsmostoftheexistingstate-of-the-artmethods, and the Bilingual Evaluation Understudy(BLEU) scores are improved with our novel method.
Zhichao Cui, yaochen Li, Chi Zhang, Yuehu Liu and Fuji Ren : Coarse-to-Fine 3D Road Model Registration for Traffic Video Augmentation, IET Image Processing, 2020.
Fuji Ren and Yanwei Bao : A Review on Human-Computer Interaction and Intelligent Robots, International Journal of Information Technology & Decision Making, Vol.19, No.1, 5-47, 2020.
(Summary)
In the field of artificial intelligence, human computer interaction (HCI) technology and its related intelligent robot technologies are essential and interesting contents of research. From the perspective of software algorithm and hardware system, these abovementioned technologies study and try to build a natural HCI environment. The purpose of this research is to provide an overview of HCI and intelligent robots. This research highlights the existing technologies of listening, speaking, reading, writing, and other senses, which are widely used in human interaction. Based on these same technologies, this research introduces some intelligent robot systems and platforms. This paper also forecasts some vital challenges of researching HCI and intelligent robots. The authors hope that this work will help researchers in the field to acquire the necessary information and technologies to further conduct more advanced research.
Fei Wang, Quan Wang, Feiping Ne, Zhongheng Li, Weizhong Yu and Fuji Ren : A linear multivariate binary decision tree classifier based on K-means splitting, Pattern Recognition, Vol.107, No.3, 1-13, 2020.
(Summary)
A novel linear multivariate decision tree classifier, Binary Decision Tree based on K-means Splitting (BDTKS), is presented in this paper. The unsupervised K-means clustering is recursively integrated into the binary tree, building a hierarchical classifier. The introduction of the unsupervised K-means clustering provides the powerful generalization ability for the resulting BDTKS model. Then, the good generalization ability of BDTKS ensures the classification performance. A novel non-split condition with an easy-setting hyperparameter which focuses more on minority classes of the current node is proposed and applied in the BDTKS model, avoiding ignoring the minority classes in the class imbalance cases. Furthermore, the K-means centroid based BDTKS model is converted into the hyperplane based decision tree, speeding up the process of classification. Extensive experiments on the publicly available data sets have demonstrated that the proposed BDTKS matches or outperforms the previous decision trees.
wenjie Liu, Wu Guoqing, Fuji Ren and XIN KANG : DFF-ResNet: An Insect Pest Recognition Model based on Residual Networks, Big Data Mining and Analytics, Vol.3, No.4, 300-310, 2020.
Jiawen Deng and Fuji Ren : Multi-label Emotion Detection via Emotion-Specified Feature Extraction and Emotion Correlation Learning, IEEE Transactions on Affective Computing, Vol.8, No.1, 2020.
(Summary)
Textual emotion detection is an attractive task while previous studies mainly focused on polarity or single-emotion classification. However, human expressions are complex, and multiple emotions often co-occur with non-negligible emotion correlations. In this paper, a Multi-label Emotion Detection Architecture (MEDA) is proposed to detect all associated emotions expressed in a given piece of text. MEDA is mainly composed of two modules: Multi-Channel Emotion-Specified Feature Extractor (MC-ESFE) and Emotion Correlation Learner (ECorL). MEDA captures underlying emotion-specified features through MC-ESFE module, which is composed of multiple channel-wise ESFE networks. Each channel in MC-ESFE is devoted to the feature extraction of a specified emotion from sentence-level to context-level through a hierarchical structure. With underlying features, emotion correlation learning is implemented through an emotion sequence predictor in ECorL. Furthermore, we define a new loss function: multi-label focal loss. With this loss function, the model can focus more on misclassified positive-negative emotion pairs and improve the overall performance by balancing the prediction of positive and negative emotions. The evaluation of proposed MEDA architecture is carried out on emotional corpus: RenCECps and NLPCC2018 datasets. The experimental results indicate that the proposed method can achieve better performance than state-of-the-art methods in this task.
Yindong Dong and Fuji Ren : Multi-reservoirs EEG signal feature sensing and recognition method based on generative adversarial networks, Computer Communications, Vol.164, 177-184, 2020.
(Summary)
EEG emotion recognition is one of the interesting and challenging tasks in the research based emotion human computer interface system. In this paper, a multi-reservoirs feature coding continuous label fusion semi-supervised Generative Adversarial Networks (MCLFS-GAN) is proposed by using permutation phase transfer entropy as the EEG signal feature. Firstly, the obtained features are encapsulated in time series, and then the features are sent into multi-reservoirs according to the division of brain intra, brain interval or frequency band. After convolution optimization, the feature expression with time sequence relationship is obtained. The generic representation between the features and pseudo effective feature expression are iteratively learned in encoder E and generator G in the generative adversarial way. In addition, the continuous fusion for class intra tags can help to form continuous differences between classes. The experimental results show that the accuracy for the four classification is 81.32% and 54.87% respectively by using SAP and LOSO in DEAP database. Compared with other models, this algorithm can effectively improve the recognition performance.
Zhong Huang, Fuji Ren, Min Hu and Sugen Chen : Facial Expression Imitation Method for Humanoid Robot Based on Smooth-Constraint Reversed Mechanical Model (SRMM), IEEE Transactions on Human-Machine Systems, Vol.50, No.6, 538-549, 2020.
(Summary)
To improve the space time similarity and motion smoothness of facial expression imitation (FEI), a real-time FEI method for a humanoid robot is proposed based on smooth-constraint reversed mechanical model (SRMM) by combining a sequence-to-sequence deep learning model and a motion-smoothing constraint. First, on the basis of facial data from a Kinect capture device, a facial feature vector is characterized based on 3 head postures, 17 facial animation units, and facial geometric deformation cascaded by Laplace coordinates. Second, a reversed mechanical model is constructed via a multilayer long short-term memory neural network to accomplish direct mapping from facial feature sequences to motor position sequences. Additionally, to overcome the motor chattering phenomenon during real-time FEI, a high-order polynomial is constructed to fit the position sequence of motors, and an SRMM is proposed and designed based on the deviation of position, velocity, and acceleration. Finally, aiming to imitate the real-time facial feature sequences of a performer captured from Kinect, the optimal position sequences generated based on the SRMM is sent to the hardware system to keep the space time characteristics consistent with those of the performer. The experimental results demonstrate that the motor position deviation of the SRMM is less than 8%. The space time similarity between the robot and the performer is greater than 85%, and the motion smoothness of the online FEI exceeded 90%. Compared with other related methods, the proposed method achieves a remarkable improvement in motor position deviation, space time similarity, and motion smoothness.
Tian Chen, Sihang Ju, Fuji Ren, Minhyan Fan and Yu Gu : EEG emotion recognition model based on the LIBSVM classifier, Measurement, Vol.164, 1-7, 2020.
(Summary)
This paper proposes an electroencephalogram(EEG) emotion recognition method based on the LIBSVM classifier. EEG features are calculated to represent the characterisitics associated with emotion states. First calculating the Lempel Ziv complexity and wavelet detail coefficients for the pre-processed EEG signals; then obtaining the co-integration relationship that reflecting the relationship between channels according to the cointegration test; next performing Empirical Mode Decomposition(EMD) on the pre-processed EEG signals; finally calculating the average approximate entropy of the first four Intrinsic Mode Functions(IMFs). The calculated four features are input into the LIBSVM classifier to realize the sentiment classification of each channel data, and then the classification results of each channel are fused by the Takagi Sugeno fuzzy model to achieve the final emotion classification. The experimental results show that when the two-category classifications are performed on Arousal and Valance, the average sentiment recognition rates are 74.88% and 82.63%, respectively.
Degen Huang, Anil Ahmed, Yasser Syed Arafat, Iftekhar Khawaja Rashid, Qasim Abbas and Fuji Ren : Sentence-Embedding and Similarity via Hybrid Bidirectional-LSTM and CNN Utilizing Weighted-Pooling Attention, IEICE Transactions on Information and Systems, Vol.E103-D, No.10, 2216-2227, 2020.
(Summary)
Neural networks have received considerable attention in sentence similarity measuring systems due to their efficiency in dealing with semantic composition. However, existing neural network methods are not sufficiently effective in capturing the most significant semantic information buried in an input. To address this problem, a novel weighted-pooling attention layer is proposed to retain the most remarkable attention vector. It has already been established that long short-term memory and a convolution neural network have a strong ability to accumulate enriched patterns of whole sentence semantic representation. First, a sentence representation is generated by employing a siamese structure based on bidirectional long short-term memory and a convolutional neural network. Subsequently, a weighted-pooling attention layer is applied to obtain an attention vector. Finally, the attention vector pair information is leveraged to calculate the score of sentence similarity. An amalgamation of both, bidirectional long short-term memory and a convolutional neural network has resulted in a model that enhances information extracting and learning capacity. Investigations show that the proposed method outperforms the state-of-the-art approaches to datasets for two tasks, namely semantic relatedness and Microsoft research paraphrase identification. The new model improves the learning capability and also boosts the similarity accuracy as well.
Jiaqiao Zhang, XIN KANG, Hongjun Ni and Fuji Ren : Surface defect detection of steel strips based on classification priority YOLOv3-dense network, Ironmaking and Steelmaking, Vol.8, 1-12, 2020.
(Summary)
The steel strip is an essential raw material in the machinery industry. Besides, the surface defects of the steel strip directly determine its performance. To achieve rapid and effective detection of the defects, a CP-YOLOv3-dense (classification priority YOLOv3 DenseNet) neural network was proposed in the present study. The model used YOLOv3 as basic network, implemented priority classification on the images, and then replaced the two residual network modules with two dense network modules. Therefore, the network can receive the multi-layer convolution features output by the dense connection block before making predictions, consequently enhancing the reuse and fusion of features. Finally, the six kinds of surface defects were detected by the improved network. According to the results, the detection precision of the CP-YOLOv3-dense network is 85.7%, the recall rate is 82.3%, the mean average precision is 82.73%, and the detection time of each image is 9.68 ms.
(Keyword)
Steel strip / defect detection / deep learning / neural network model / surface technology
XIAOHUA WANG, JIANQIAO GONG, MIN HU, YU GU and Fuji Ren : LAUN Improved StarGAN for Facial Emotion Recognition, IEEE Access, Vol.8, No.1, 161509-161518, 2020.
(Summary)
In the eld of facial expression recognition, deep learning is extensively used. However, insuf cient and unbalanced facial training data in available public databases is a major challenge for improving the expression recognition rate. Generative Adversarial Networks (GANs) can produce more one-to-onefaceswithdifferentexpressions,whichcanbeusedtoenhancedatabases.StarGANcanperform one-to-many translations for multiple expressions. Compared with original GANs, StarGAN can increase the ef ciency of sample generation. Nevertheless, there are some defects in essential areas of the generated face,suchasthemouthandthefuzzysidefaceimagegeneration.Toaddresstheselimitations,weimproved StarGAN to alleviate the defects of images generation by modifying the reconstruction loss and adding the Contextual loss. Meanwhile, we added the Attention U-Net to StarGAN's generator, replacing StarGAN's original generator. Therefore, we proposed the Contextual loss and Attention U-Net (LAUN) improved StarGAN.TheU-shapestructureandskipconnectioninAttentionU-Netcaneffectivelyintegratethedetails and semantic features of images. The network's attention structure can pay attention to the essential areas of the human face. The experimental results demonstrate that the improved model can alleviate some aws in the face generated by the original StarGAN. Therefore, it can generate person images with better quality withdifferentposesandexpressions.TheexperimentswereconductedontheKarolinskaDirectedEmotional Facesdatabase,andtheaccuracyoffacialexpressionrecognitionis95.97%,2.19%higherthanthatbyusing StarGAN. Meanwhile, the experiments were carried out on the MMI Facial Expression Database, and the accuracy of expression is 98.30%, 1.21% higher than that by using StarGAN. Moreover, experiment results have better performance based on the LAUN improved StarGAN enhanced databases than those without enhancement.
Yangyang Zhou and Fuji Ren : CERG: Chinese Emotional Response Generator with Retrieval Method, Research : A Science Partner Journal, Vol.2020, No.1, 1-8, 2020.
(Summary)
The dialogue system has always been one of the important topics in the domain of artificial intelligence. So far, most of the mature dialogue systems are task-oriented based, while non-task-oriented dialogue systems still have a lot of room for improvement. We propose a data-driven non-task-oriented dialogue generator ``CERG'' based on neural networks. This model has the emotion recognition capability and can generate corresponding responses. The data set we adopt comes from the NTCIR-14 STC-3 CECG subtask, which contains more than 1.7 million Chinese Weibo post-response pairs and 6 emotion categories. We try to concatenate the post and the response with the emotion, then mask the response part of the input text character by character to emulate the encoder-decoder framework. We use the improved transformer blocks as the core to build the model and add regularization methods to alleviate the problems of overcorrection and exposure bias. We introduce the retrieval method to the inference process to improve the semantic relevance of generated responses. The results of the manual evaluation show that our proposed model can make different responses to different emotions to improve the human-computer interaction experience. This model can be applied to lots of domains, such as automatic reply robots of social application.
Jun Liu, Xiuyun Wang and Fuji Ren : Test Strategy for 2.5D ICs Using Auxiliary Interposer and E-fuses, Computer Engineering and Applications, Vol.39, No.6, 1863-1868, 2020.
Zheng Liu, XIN KANG, Shun Nishide and Fuji Ren : Vowel priority lip matching scheme and similarity evaluation model based on humanoid robot Ren-Xin, Journal of Ambient Intelligence and Humanized Computing, 2020.
Leyuan Jia, Yu Gu, Ken Cheng, Huan Yan and Fuji Ren : BeAware: Convolutional neural network(CNN) based user behavior understanding through WiFi channel state information, Neurocomputing, Vol.397, 457-463, 2020.
(Summary)
In modern informatics society, human beings are becoming more and more attached to the computer. Therefore, understanding user behavior is critical to various application elds like sedentary analysis, human-computer interaction, and affective computing. Current sensor-based and vision-based user behavior understanding approaches are either contact or obtrusive to users, jeopardizing their availability and practicality. To this end, we present BeAware, a contactless Radio Frequency (RF) based user behavior understanding system leveraging the WiFi Channel State Information (CSI). The key idea is to visualize the channel data affected by human movements into time-series heat-map images, which are processed by a Convolutional Neural Network (CNN) to understand the corresponding user behaviors. We prototype BeAware on commodity low-cost WiFi devices and evaluate its performance in real-world environments. Experimental results have veri ed its effectiveness in recognizing user behaviors.
Yu Gu, Yantong Wang, Tao Liu, Yusheng Ji, Zhi Liu, Peng Li, Xiaoyan Wang, Xin Ai and Fuji Ren : EmoSense: Computational Intelligence Driven Emotion Sensing via Wireless Channel Data, IEEE Transactions on Emerging Topics in Computational Intelligence, Vol.4, No.3, 216-226, 2020.
(Summary)
Emotion is well recognized as a distinguished symbol ofhumanbeings,anditplaysacrucialroleinourdailylives.Existingvision-basedorsensor-basedsolutionsareeitherobstructiveto use or rely on specialized hardware, hindering their applicability. This paper introduces EmoSense, a rst-of-its-kind wireless emotionsensingsystemdrivenbycomputationalintelligence.Thebasic methodologyistoexplorethephysicalexpressionofemotionsfrom wireless channel response via data mining. The design and implementation of EmoSense faces two major challenges extracting physical expression from wireless channel data and recovering emotion from the corresponding physical expression. For the former, we present a Fresnel zone-based theoretical model depicting the ngerprint of the physical expression on channel response. For the latter, we design an ef cient computational intelligence driven mechanism to recognize emotion from the corresponding ngerprints.WeprototypedEmoSenseonthecommodityWiFiinfrastructure and compared it with mainstream sensor-based and vision-basedapproachesinthereal-worldscenario.Thenumerical study over 3360 cases con rms that EmoSense achieves a comparable performance to the vision-based and sensor-based rivals under different scenarios. EmoSense only leverages the low-cost and prevalent WiFi infrastructures and thus, constitutes a tempting solution for emotion sensing.
Yu Gu, Jinhai Zhan, Jie Li, Yusheng Ji, Xin An and Fuji Ren : Sleepy: Wireless Channel Data Driven Sleep Monitoring via Commodity WiFi Devices, IEEE Transactions on Big Data, Vol.6, No.2, 258-268, 2020.
(Summary)
Sleep is a major event of our daily lives. Its quality constitutes a critical indicator of people's health conditions, both mentally and physically. Existing sensor-based or vision-based sleep monitoring systems either are obstructive to use or fail to provide adequate coverage. With the fast expansion of wireless infrastructures nowadays, channel data, which is pervasive and transparent, emerges as another alternative. To this end, we propose Sleepy, a wireless channel data driven sleep monitoring system leveraging commercial WiFi devices. The key idea of Sleepy is that the energy feature of the wireless channel follows a Gaussian Mixture Model (GMM) derived from the accumulated channel data over a long period. Therefore, a GMM based foreground extraction method has been designed to adaptively distinguish motions like rollovers (foreground) from background (stationary postures), leading to certain major merits, e.g., no calibrations or target-dependent training needed. We prototype Sleepy and evaluate it in two real environments. In the short-term controlled experiments, Sleepy achieves 95.04% detection accuracy and 4.07% false negative rate. In the 60-minute real case studies, Sleepy demonstrates strong stability. Considering that Sleepy is compatible with existing WiFi infrastructure, it constitutes a low-cost yet promising solution for sleep monitoring.
Fuji Ren and Siyuan Xue : Intention Detection Based on Siamese Neural Network With Triplet Loss, IEEE Access, Vol.8, No.1, 82242-82254, 2020.
(Summary)
Understanding the user's intention is an essential task for the spoken language understanding (SLU) module in the dialogue system, which further illustrates vital information for managing and generating future action and response. In this paper, we propose a triplet training framework based on the multiclassclassi cationapproachtoconductthetrainingfortheintentiondetectiontask.Precisely,weutilize aSiameseneuralnetworkarchitecturewithmetriclearningtoconstructarobustanddiscriminativeutterance feature embedding model. We modi ed the RMCNN model and ne-tuned BERT model as Siamese encoderstotrainutterancetripletsfromdifferentsemanticaspects.Thetripletlosscaneffectivelydistinguish the details of two input data by learning a mapping from sequence utterances to a compact Euclidean space. After generating the mapping, the intention detection task can be easily implemented using standard techniques with pre-trained embeddings as feature vectors. Besides, we use the fusion strategy to enhance utterance feature representation in the downstream of intention detection task. We conduct experiments on several benchmark datasets of intention detection task: Snips dataset, ATIS dataset, Facebook multilingual task-oriented datasets, Daily Dialogue dataset, and MRDA dataset. The results illustrate that the proposed methodcaneffectivelyimprovetherecognitionperformanceofthesedatasetsandachievesnewstate-of-theart results on single-turn task-oriented datasets (Snips dataset, Facebook dataset), and a multi-turn dataset (Daily Dialogue dataset).
Siyuan Xue and Fuji Ren : Knowledge Graph based Question and Answer System for Cosmetic Domain, International Journal of Advanced Intelligence (IJAI), Vol.11, No.1, 15-28, 2020.
Fuji Ren and Qian Zhang : An Emotion Expression Extraction Method for Chinese Microblog Sentences, IEEE Access, Vol.8, No.1, 69244-69255, 2020.
(Summary)
With the rapid spread of Chinese microblog, a large number of microblog topics are being generated in real-time. More and more users pay attention to emotion expressions of these opinionated sentences in different topics. It is challenging to label the emotion expressions of opinionated sentences manually. For this endeavor, an emotion expression extraction method is proposed to process millions of user-generated opinionated sentences automatically in this paper. Specifically, the proposed method mainly contains two tasks: emotion classification and opinion target extraction. We first use a lexicon-based emotion classification method to compute different emotion values in emotion label vectors of opinionated sentences. Then emotion label vectors of opinionated sentences are revised by an unsupervised emotion label propagation algorithm. After extracting candidate opinion targets of opinionated sentences, the opinion target extraction task is performed on a random walk-based ranking algorithm, which considers the connection between candidate opinion targets and the textual similarity between opinionated sentences, ranks candidate opinion targets of opinionated sentences. Experimental results demonstrate the effectiveness of algorithms in the proposed method.
Yu Zhao, Anxiang Zhang, Huali Feng, Qing Li, Patrick Gallinari and Fuji Ren : Knowledge graph entity typing via learning connecting embeddings, Knowledge-based Systems, No.1, 2020.
(Summary)
Knowledge graph (KG) entity typing aims at inferring possible missing entity type instances in KG, which is a very significant but still under-explored subtask of knowledge graph completion. In this paper, we propose a novel approach for KG entity typing which is trained by jointly utilizing local typing knowledge from existing entity type assertions and global triple knowledge in KGs. Specifically, we present two distinct knowledge-driven effective mechanisms of entity type inference. Accordingly, we build two novel embedding models to realize the mechanisms. Afterward, a joint model via connecting them is used to infer missing entity type instances, which favors inferences that agree with both entity type instances and triple knowledge in KGs. Experimental results on two real-world datasets (Freebase and YAGO) demonstrate the effectiveness of our proposed mechanisms and models for improving KG entity typing.
Fuji Ren and Yangyang Zhou : CGMVQA: A New Classification and Generative Model for Medical Visual Question Answering, IEEE Access, Vol.8, No.1, 50626-50636, 2020.
(Summary)
Medical images are playing an important role in the medical domain. A mature medical visual question answering system can aid diagnosis, but there is no satisfactory method to solve this comprehensive problem so far. Considering that there are many different types of questions, we propose a model called CGMVQA, including classification and answer generation capabilities to turn this complex problem into multiple simple problems in this paper. We adopt data augmentation on images and tokenization on texts. We use pre-trained ResNet152 to extract image features and add three kinds of embeddings together to deal with texts. We reduce the parameters of the multi-head self-attention transformer to cut the computational cost down. We adjust the masking and output layers to change the functions of the model. This model establishes new state-of-the-art results: 0.640 of classification accuracy, 0.659 of word matching and 0.678 of semantic similarity in ImageCLEF 2019 VQA-Med data set. It suggests that the CGMVQA is effective in medical visual question answering and can better assist doctors in clinical analysis and diagnosis.
Xiuyun Wang, Jun Liu and Fuji Ren : TSVs crosstalk fault grouping test and diagnostic strategy, MICROELECTRONICS & COMPUTER, Vol.37, No.2, 30-36, 2020.
Tian Chen, Yang Zhou, Fuji Ren, Xin An and Huyin Zhao : Improved run-length codes compression method based on tri-state signal, Computer Engineering, 1-8, 2020.
OMAR AHMED, Fuji Ren, AMMAR HAWBANI and YASER AL-SHARABI : Energy Optimized Congestion Control Based Temperature Aware Routing Algorithm For Software Defined Wireless Body Area Networks, IEEE Access, Vol.8, 41085-41099, 2020.
(Summary)
Wireless Body Area Network(WBAN) is a promising cost-effective technology for the privacy con ned military applications and healthcare applications like remote health monitoring, telemedicine, and e-health services. The use of a Software-De ned Network (SDN) approach improves the control and management processes of the complex structured WBANs and also provides higher exibility and dynamic network structure. To seamless routing performance in SDN-based WBAN, the energy-ef ciency problems must be tackled effectively. The main contribution of this paper is to develop a novel Energy Optimized Congestion Control based on Temperature Aware Routing Algorithm (EOCC-TARA) using Enhanced Multi-objective Spider Monkey Optimization (EMSMO) for SDN-based WBAN. This algorithm overcomes the vital challenges, namely energy-ef ciency, congestion-free communication, and reducing adverse thermal effectsin WBAN routing. First,the proposed EOCC-TARA routing algorithm considers the effects of temperature due to the thermal dissipation of sensor nodes and formulates a strategy to adaptively select the forwarding nodes based on temperature and energy. Then the congestion avoidance concept is added with the energy-ef ciency, link reliability, and path loss for modeling the cost function based on which the EMSMO provides the optimal routing. Simulations were performed, and the evaluation results showed that the proposed EOCC-TARA routing algorithm has superior performance than the traditional routing approaches in terms of energy consumption, network lifetime, throughput, temperature control, congestion overhead, delay, and successful transmission rate.
Xiaohua Wang, Lijuan Pan, Muzi Peng, Min Hu, Chunhua Jin and Fuji Ren : Video Emotion Recognition Based on Hierarchical Attention Model, Journal of Computer-Aided Design & Computer Graphics, Vol.32, No.1, 27-35, 2020.
(Summary)
LSTM network is widely used in facial expression recognition of video sequences. In view of the limited representation ability of single-layer LSTM and the limitation of its generalization ability when solving complex problems, a hierarchical attention model is proposed. Hierarchical representation of time series data is learned by stacking LSTM, self-attention mechanism is used to construct differentiated hierarchical relationships, and a penalty term is constructed and further combined with the loss function to optimize the network performance. Experiments on CK+ and MMI datasets, demonstrate that due to the construction of good hierarchical features, each step in time series can select information from the more interesting feature hierarchy. Compared with ordinary single-layer LSTM, hierarchical attention model can express the emotional information of video sequences more effectively.
Zhichao Cui, Yuehu Liu and Fuji Ren : Homography-based traffic sign localisation and pose estimation from image sequence, IET Image Processing, Vol.13, No.14, 2829-2839, 2019.
(Summary)
This study proposes a vision-based method for traffic sign attribute estimation, i.e. 3D position and pose, from image sequences by binocular or monocular cameras. The method starts with acquiring robust feature correspondences based on homography constraints from image pairs. Then the objective function is designed to integrate the feature correspondences to optimise the parameters of the traffic sign plane in the 3D coordinate. Finally, the sign plane is utilised for attribute estimation. In addition, the authors provide an extension for the raw KITTI dataset, which can be utilised for 3D tasks of traffic sign localisation and pose estimation. In the experiments, three popular methods are employed for comparisons based on the publicly available BelgiumTS and KITTI datasets. The results show that the authors' method based on SIFT and SURF features can locate the traffic signs with a mean error of ∼0.44 and 0.51 m in the BelgiumTS and KITTI datasets, respectively, and estimate the pose with a mean error of ∼14.45° in the KITTI dataset.
Hongjun Ni, Jiaqiao Zhang, Nansheng Zhao, Chusen Wang, Shuaishuai Lv, Fuji Ren and Xingxing Wang : Design on the Winter Jujubes Harvesting and Sorting Device, Applied Sciences, No.1, 1-17, 2019.
(Summary)
According to the existing problems of winter jujube harvesting, such as the intensive labor of manual picking, damage to the surface of winter jujubes, a winter jujube harvesting and sorting device was developed. This device consisted of vibration mechanism, collection mechanism, and sorting mechanism. The eccentric vibration mechanism made the winter jujubes fall, and the umbrella collecting mechanism can collect winter jujube and avoid the impact of winter jujube on the ground, and the sorting mechanism removed jujube leaves and divided the jujube into two types, and the automatic leveling mechanism made the device run smoothly in the field. Through finite element analysis and BP (Back Propagation) neural network analysis, the results show that: The vibration displacement of jujube tree is related to the trunk diameter and vibration position; the impact force of winter jujubes falling is related to the elastic modulus of umbrella material; the collecting area can be increased four times for each additional step of the collection mechanism; jujube leaves can be effectively removed when blower wind speed reaches 45.64 m/s. According to the evaluation standard grades of the jujubes harvesting and sorting, the device has good effects and the excellent rate up to 90%, which has good practicability and economy.
Tian Chen, Yongsheng Zuo, Xin An and Fuji Ren : A variable-length compatible compression scheme based on tristate signals, The Journal of Supercomputing, 1-14, 2019.
(Summary)
The test data volume for the manufacturing test of chips is increasing rapidly. This is due to the fact that the complexity of these chips is increasing rapidly and the transistor count is increasing exponentially. Aimed at solving the problems of large test data volume and long test time in chips test, this paper presents a deterministic test-based compression method that uses tristate signals encoding and compatible test cube merging. Firstly, the partial input reduction is carried out for the original test set. The proportion of ``don't care'' bits in the original test set is increased by merging compatible or inversely compatible inputs in the test cubes, so the compatibility among the test cubes is improved. Then, the test set is compressed and encoded by using tristate signal. Each test cube is divided into several data segments, which are encoded by tristate signals and multiple compatibility rules. This scheme can improve the compression ratio of the test set. The experimental results show that the proposed scheme achieves a good compression ratio, without excessive test power and area overhead. The average test compression ratio can reach 82.15%.
Jun Liu, Peng Dong and Fuji Ren : Diagonal-based for TSV Fault Tolerance Design, Microelectronics & Computer, 2019.
(Summary)
Three dimensional integrated circuits, using through-silicon vias(TSVs) as the communication link between vertical chip, with the advantages of high density, high bandwidth and low power consumption. As the TSV during processing and using process may be failure, will lead to the entire three-dimensional chip failure. In order to improve the yield of 3D chips, the yield of TSV must be raised as much as possible. This paper presents a novel diagonal-based redundant TSV repair scheme and proposes a repair algorithm based on maximum flow algorithm to tolerance the faulty TSV with redundant TSVs in the diagonal to improve the yield of the entire three-dimensional chip. Experimental results show that the yield of the proposed repair strategy is 98.38% to 98.96% and the hardware overhead can be reduced by up to 70% compared to routerbased technique.
71.
Xinyu He, Lishuang Li, Xingchen Song, Degen Huang and Fuji Ren : Multi-Level Attention Based BLSTM Neural Network for Biomedical Event Extraction, IEICE Transactions on Information and Systems, Vol.E102-D, No.9, 1842-1850, 2019.
(Summary)
Biomedical event extraction is an important and challenging task in Information Extraction, which plays a key role for medicine research and disease prevention. Most of the existing event detection methods are based on shallow machine learning methods which mainly rely on domain knowledge and elaborately designed features. Another challenge is that some crucial information as well as the interactions among words or arguments may be ignored since most works treat words and sentences equally. Therefore, we employ a Bidirectional Long Short Term Memory (BLSTM) neural network for event extraction, which can skip handcrafted complex feature extraction. Furthermore, we propose a multi-level attention mechanism, including word level attention which determines the importance of words in a sentence, and the sentence level attention which determines the importance of relevant arguments. Finally, we train dependency word embeddings and add sentence vectors to enrich semantic information. The experimental results show that our model achieves an F-score of 59.61% on the commonly used dataset (MLEE) of biomedical event extraction, which outperforms other state-of-the-art methods.
Fuji Ren, WenJie Liu and Guoqing Wu : Feature Reuse Residual Networks for Insect Pest Recognition, IEEE Access, Vol.7, 122758-122768, 2019.
(Summary)
Insect pests are one of the main threats to the commercially important crops. An effective insect pest recognition method can avoid economic losses. In this paper, we proposed a new and simple structure based on the original residual block and named as feature reuse residual block which combines feature from the input signal of a residual block with the residual signal. In each feature reuse residual block, it enhances the capacity of representation by learning half and reuse half feature. By stacking the feature reuse residual block, we obtained the feature reuse residual network (FR-ResNet) and evaluated the performance on IP102 benchmark dataset. The experimental results showed that FR-ResNet can achieve significant performance improvement in terms of insect pest classification. Moreover, to demonstrate the adaptive of our approach, we applied it to various kinds of residual networks, including ResNet, Pre-ResNet, and WRN, and we tested the performance on a series of benchmark datasets: CIFAR-10, CIFAR-100, and SVHN. The experimental results showed that the performance can be improved obviously than original networks. Based on these experiments on CIFAR-10, CIFAR-100, SVHN, and IP102 benchmark datasets, it demonstrates the effectiveness of our approach.
MIN HU, CHUNJIAN YANG, YAQIN ZHENG, XIAOHUA WANG, LEI HE and Fuji Ren : Facial Expression Recognition Based on Fusion Features of Center-Symmetric Local Signal Magnitude Pattern, IEEE Access, Vol.7, No.1, 118435-118445, 2019.
(Summary)
Local feature descriptors play a fundamental and important role in facial expression recognition. This paper presents a new descriptor, Center-Symmetric Local Signal Magnitude Pattern (CS-LSMP), which is used for extracting texture features from facial images. CS-LSMP operator takes signal and magnitude information of local regions into account compared to conventional LBP-based operators. Additionally, due to the limitation of single feature extraction method and in order to make full advantagesofdifferentfeatures,thispaperemploysCS-LSMPoperatortoextractfeaturesfromOrientational Magnitude Feature Maps (OMFMs), Positive-and-Negative Magnitude Feature Maps (PNMFMs), Gabor FeatureMaps(GFMs)andfacialpatches(eyebrows-eyes,mouths)forobtainingfusedfeatures.UnlikeHOG, which only retains horizontal and vertical magnitudes, our work generates Orientational Magnitude Feature Maps (OMFMs) by expanding multi-orientations. This paper build two distinct feature maps by dividing local magnitudes into two groups, i.e., positive and negative magnitude feature maps. The generated Gabor FeatureMaps(GFMs)arealsogroupedtoreducethecomputationalcomplexity.ExperimentsontheJAFFE andCK+facialexpressiondatasetsshowedthattheproposedframeworkachievedsigni cantimprovement and outperformed some state-of-the-art methods.
MIN HU, YAQIN ZHANG, CHUNJIAN YANG, XIAOHUA WANG, LEI HE, Fuji Ren and XIAOYIN XU : Facial Expression Recognition Using Fusion Features Based on Center-Symmetric Local Octonary Pattern, IEEE Access, Vol.7, No.1, 29882-29890, 2019.
(Summary)
A local feature descriptor has gained a lot of interest in many applications, such as image retrieval, texture classi cation, and face recognition. This paper proposes a novel local feature descriptor: center-symmetric local octonary pattern (CS-LOP) for facial expression recognition. A CS-LOP operator not only considers the difference of the gray value between central pixels and neighboring pixels in all eight directions but also compares the gray value of four pairs of center-symmetric pixels. Besides, this paper used the CS-LOP to extract diverse features from the preprocessed facial image, the feature map of gradient magnitude, and the feature map of Gabor, and also to make extracted features more abundant and detailed. To evaluate the performance of the proposed method, experiments on JAFFE and CK facial expression datasets demonstrate that the proposed method outperforms the method using the individual descriptor. Compared with other state-of-the-art methods, our approach improves the overall recognition accuracy.
Yu Gu, Xiang Zhang, Zhi Liu and Fuji Ren : BeSense: Leveraging WiFi Channel Data and Computational Intelligence for Behavior Analysis, IEEE Computational Intelligence Magazine, Vol.14, No.4, 31-41, 2019.
(Summary)
The ever-evolving informatics technology has gradually bounded human and computer in a compact way. Understanding user behavior becomes a key enabler in many fields such as sedentary-related healthcare, human-computer interaction and affective computing. Traditional sensor-based and vision-based user behavior analysis approaches are obtrusive in general, hindering their usage in real-world. Therefore, in this article, we first introduce the WiFi signal as a new source instead of sensor and vision for unobtrusive user behaviors analysis. Then we design BeSense, a contactless behavior analysis system leveraging signal processing and computational intelligence over WiFi channel state information. We prototype BeSense on commodity low-cost WiFi devices and evaluate its performance in real-world environments. Experimental results have verified its effectiveness in recognizing user behaviors.
JUAN LIU, ZHONG HUANG, Fuji Ren and LEI HUA : Drug-Drug Interaction Extraction Based on Transfer Weight Matrix and Memory Network, IEEE Access, Vol.7, No.1, 101260-101268, 2019.
(Summary)
Extracting drug drug interaction (DDI) in the text is the process of identifying how two target drugs in a given sentence interact. Previous methods, which were limited to conventional machine learning techniques, we are susceptible to issues such as ``vocabulary gap'' and unattainable automation processes in feature extraction. Inspired by deep learning in natural language preprocessing, we addressed theaforementionedproblemsbasedondynamictransfermatrixandmemorynetworks.ATM-RNNmethod is proposed by adding the transfer weight matrix in multilayer bidirectional LSTM to improve robustness andintroduceamemorynetworkforfeaturefusion.WeevaluatedtheTM-RNNmodelontheDDIExtraction 2013Task.TheproposedmodelachievedanoverallF-scoreof72.43,whichoutperformsthelatestmethods based on support vector machine and other neural networks. Meanwhile, the experimental results also indicated that the proposed model is more stable and less affected by negative samples.
(Keyword)
Drug drug interaction extraction / memory network / multilayer bidirectional LSTM
Tian Chen, Yongsheng Zuo, Xin An and Fuji Ren : Test data compatible compression method based on tri-state signal, Journal of Computer Applications, Vol.39, No.6, 1863-1868, 2019.
Xiaohua Wang, Muzi Peng, Lijuan Pan, Min Hu, Chunhua Jin and Fuji Ren : Two-level attention with two-stage multi-task learning for facial emotion recognition, Journal of Visual Communication and Image Representation, Vol.62, No.1, 217-225, 2019.
(Summary)
Compared with facial emotion recognition on categorical model, the dimensional emotion recognition can describe numerous emotions of the real world more accurately. Most prior works of dimensional emotion estimation only considered laboratory data and used video, speech or other multi-modal features. The effect of these methods applied on static images in the real world is unknown. In this paper, a two-level attention with two-stage multi-task learning (2Att-2Mt) framework is proposed for facial emotion estimation on only static images. Firstly, the features of corresponding region(position-level features) are extracted and enhanced automatically by first-level attention mechanism. In the following, we utilize Bi-directional Recurrent Neural Network(Bi-RNN) with self-attention(second-level attention) to make full use of the relationship features of different layers(layer-level features) adaptively. Owing to the inherent complexity of dimensional emotion recognition, we propose a two-stage multi-task learning structure to exploited categorical representations to ameliorate the dimensional representations and estimate valence and arousal simultaneously in view of the correlation of the two targets. The quantitative results conducted on AffectNet dataset show significant advancement on Concordance Correlation Coefficient(CCC) and Root Mean Square Error(RMSE), illustrating the superiority of the proposed framework. Besides, extensive comparative experiments have also fully demonstrated the effectiveness of different components.
Tian Chen, Zhangang Chen, Xiaohui Yuan, Sihang Ju and Fuji Ren : Emotion Recognition Method Based on Instantaneous Energy of Electroencephalography, Computer Engineering, Vol.45, No.4, 196-204, 2019.
Xiao Sun, Tao Hong, Changliang Li and Fuji Ren : Hybrid Spatiotemporal Models for Sentiment Classification Via Galvanic Skin Response, Neurocomputing, Vol.358, 385-400, 2019.
(Summary)
Sentiment plays an important role in cognition, creativity, attention, and decision making in people's daily lives. Researchers have made great progress in sentiment recognition through images and speech. In this paper, a multimodal dataset is proposed for sentiment classi cation (MDSTC 1 ) a multimodal dataset col lected with multimodal channels by customized physiological sensors and manually labelled for human sentiment analysis. MDSTC contains galvanic skin response (GSR), pulse, speech, and facial expression data of 100 volunteers labelled with timeline, emotional self-ratings, and personal information such as age, sex and Big Five personality scores, which were collected while watching video clips to stimulate emotions. Using the MDSTC dataset, the GSR channel with six types of emotion labels is used to perform human sentiment analysis in this paper. The GSR signal is converted into a spectrogram to adopt image related methods and deep learning models. A convolutional neural network, long short-term memory, and self-attention mechanism are adopted and combined, and spatiotemporal hybrid models are pro posed to perform human sentiment analysis. Comparisons are performed between the proposed model and state-of-the-art models in related works. The experimental results show that with the proposed spa tiotemporal hybrid models, better results are obtained with respect to precision, recall, and F1-score. The experimental results also show that with the proposed spatiotemporal hybrid models working with GSR, people's emotional changes can be obtained in real time with high precision.
XIN KANG, Yunong Wu and Fuji Ren : Toward action comprehension for searching: Mining actionable intents in query entities, Journal of the Association for Information Science and Technology, 2019.
Xiaoxia Liu, Degen Huang, Zhangzhi Yin and Fuji Ren : Recognition of Collocation Frames from Sentences, IEICE Transactions on Information and Systems, Vol.E102-D, No.3, 620-627, 2019.
(Summary)
Collocation is a ubiquitous phenomenon in languages and accurate collocation recognition and extraction is of great signi cance to many natural language processing tasks. Collocations can be di erentiated from simple bigram collocations to collocation frames(referring to distant multi-gram collocations). So far little focus is put on collocation frames. Oriented to translation and parsing, this study aims to recognize and extract the longest possible collocation frames from given sentences. We rst extract bigram collocations with distributional semantics based method by introducing collocation patterns and integrating some state-of-the-art association measures. Based on bigram collocations extracted by the proposed method, we get the longest collocation frames according to recursive nature and linguistic rules of collocations. Compared with the baseline systems, the proposed method performs signi cantly better in bigram collocation extraction both inprecision and recall. And in extracting collocation frames, the proposed method performs even better with the precision similar to its bigram collocation extraction results.
XINHUA CAO, TAIHAO Li, HONGLI LI, SHUNREN XIA, Fuji Ren, YE SUN and XIAOYIN XU : A Robust Parameter-Free Thresholding Method for Image Segmentation, IEEE Access, Vol.7, No.1, 3448-3458, 2019.
(Summary)
Inthispaper,wepresentedanewparameter-freethresholdingmethodforimagesegmentation. In separating an image into two classes, the method employs an objective function that not only maximizes the between-class variance but also the distance between the mean of each class and the global mean of the image. The design of the objective function aims to circumvent the challenge that many existing techniques encounter when the underlying two classes have very different sizes or variances. The advantages of the new method are twofold. First, it is parameter-free, meaning that it can generate consistent results. Second, the new method has a simple form that makes it easy to adapt to different applications. We tested and compared the new method with the standard Otsu method, the maximum entropy method, and the 2D Otsu method on the simulated and real biomedical and photographic images and found that the new method can achieve a more accurate and robust performance.
(Keyword)
Segmentation / parameter-free thresholding / objective function / histogram
Kazuyuki Matsumoto, Fuji Ren, Masaya Matsuoka, Minoru Yoshida and Kenji Kita : Slang Feature Extraction by Analyzing Topic Change on Social Media, CAAI Transactions on Intelligence Technology, 2019.
(Summary)
Recently, we often see words such as youth slang, neologism and Internet slang on social networking sites (SNSs) that are not registered on dictionaries. Because the documents posted to SNSs include a lot of fresh information, they are thought to be useful for collecting information. It is important to analyze these words (hereinafter referred to as slang) and capture their features for the improvement of the accuracy of automatic information collection. This paper aims to analyze what features can be observed in slang by focusing on the topic. We construct topic models from document groups including target slang on Twitter by Latent Dirichlet Allocation (LDA). With the models, we chronologically analyze change of topics during a certain period of time to find out the difference in the features between slang and general words. Then, we propose a slang classification method based on the change of features.
Fuji Ren and Jiawen Deng : Background Knowledge Based Multi-Stream Neural Network for Text Classification, Applied Sciences, No.1, 1-18, 2018.
(Summary)
As a foundation and typical task in natural language processing, text classification has been widely applied in many fields. However, as the basis of text classification, most existing corpus are imbalanced and often result in the classifier tending its performance to those categories with more texts. In this paper, we propose a background knowledge based multi-stream neural network to make up for the imbalance or insufficient information caused by the limitations of training corpus. The multi-stream network mainly consists of the basal stream, which retains original sequence information, and background knowledge-based streams. Background knowledge is composed of keywords and co-occurred words which are extracted from external corpus. Background knowledge-based streams are devoted to realizing supplemental information and reinforce the basal stream. To better fuse the features extracted from different streams, early-fusion and two after-fusion strategies are employed. According to the results obtained from both Chinese corpus and English corpus, it is demonstrated that the proposed background knowledge based multi-stream neural network performs well in classification tasks.
Zhao Han, Fuji Ren and Duoqian Miao : Sentiment Analysis Method Based on an Improved Modifying-MatrixLanguage Model, IEEJ Transactions on Electrical and Electronic Engineering (TEEE), Vol.13, No.10, 1446-1453, 2018.
(Summary)
In this paper, a sentence-level sentiment analysis method is proposed to deal with sentiment measurement and classificationproblems. It is developed from a model called the synthetic and computational language model (SCLM), which representsmodifying and modified information, respectively, using matrices and vectors. In the proposed method, a global modifyingmatrix of a sentence is constructed, the determinant value of this matrix is calculated and adjusted, and then the final valueis used as the sentiment value of the sentence. Regression experiment shows that the deviation between the output sentimentand the target sentiment does not exceed a class distance of five classes. The classification experiment shows that the proposedmethod has improved most of the performance compared to the simplified SCLM and in some cases, such as in `very positive'class and `very negative' class, reaches higher precision performance than the baseline method.
(Keyword)
affective computing / sentiment analysis / emotion classification / language model / neural network
Tian Chen, Sihang Ju, Xiaohui Yuan, Mohamed Elhoseny, Fuji Ren and Mingyan Fan : Emotion recognition using empirical mode decomposition and approximation entropy, Computers & Electrical Engineering, Vol.72, No.1, 383-392, 2018.
(Summary)
Automatic human emotion recognition is a key technology for human-machine interac tion. In this paper, we propose an electroencephalogram (EEG) feature extraction method that leverages empirical mode decomposition and Approximation Entropy. In our proposed method, Empirical Mode Decomposition (EMD) is used to process EEG signals after data processing and obtains several intrinsic eigenmode functions. The Approximation Entropy (ApEn) of the rst four Intrinsic Mode Functions (IMFs) is computed, which is used as the features from EEG signals for learning and recognition. An integration of Deep Belief Net work and Support Vector Machine is devised for classi cation, which takes the eigenvec tors from the extracted feature to identify four principal human emotions, namely happy, calm, sad, and fear. Experiments are conducted with EEG data acquired with a 16-lead device. Our experimental results demonstrate that the proposed method achieves an im proved accuracy that is highly competitive to the state-of-the-art methods. The average accuracy is 83.34%, and the best accuracy reaches 87.32%
Fuji Ren, Yindong Dong and Wei Wang : Emotion recognition based on physiological signals using brain asymmetry index and echo state network, Neural Computing & Applications, 2018.
(Summary)
This paper proposes a method to evaluate the degree of emotion being motivated in continuous music videos based on asymmetry index (AsI). By collecting two groups of electroencephalogram (EEG) signals from 6 channels (Fp1, Fp2, Fz and AF3, AF4, Fz) in the left and right hemispheres, multidimensional directed information is used to measure the mutual information shared between two frontal lobes, and then, we get AsI to estimate the degree of emotional induction. In order to evaluate the effect of AsI processing on physiological emotion recognition, 32-channel EEG signals, 2-channel EEG signals and 2-channel EMG signals are selected for each subject from the DEAP dataset, and different sub-bands are extracted using wavelet packet transform. k-means algorithm is used to cluster the wavelet packet coef cients of each subband, and the probability distribution of the coef cients under each cluster is calculated. Finally, the probability distribution value of each sample is sent as the original features into echo state network for unsupervised intrinsic plasticity training; the reservoir state nodes are selected as the nal feature vector and fed into the support vector machine. The experimental results show that the proposed algorithm can achieve an average recognition rate of 70.5% when the subjects are independent. Compared with the case without AsI, the recognition rate is increased by 8.73%. On the other hand, the ESN is adopted for the original physiological feature re nement which can signi cantly reduce feature dimensions and be more bene cial to the emotion classi cation. Therefore, this study can effectively improve the performance of human machine interface systems based on emotion recognition.
Xiaohua Wang, Ying Chen, Min Hu and Fuji Ren : Occluded Facial Expression Recognition Based on Asymmetric Region Weber Local Descriptor and Block Similarity Weighting, Laser & Optoelectronics Progress, Vol.55, No.4, 1-8, 2018.
Min Hu, Keke Zhang, Xiaohua Wang and Fuji Ren : Video facial expression recognition combined with sliding window dynamic time warping and CNN, Journal of Image and Graphics, Vol.23, No.8, 1143-1153, 2018.
91.
Li Quan, Zhiliang Wang and Fuji Ren : A Novel Two-Layered Reinforcement Learning for Task Of oading with Tradeoff between Physical Machine Utilization Rate and Delay, Future Internet, Vol.60, No.10, 1-17, 2018.
(Summary)
Mobile devices could augment their ability via cloud resources in mobile cloud computing environments. This paper developed a novel two-layered reinforcement learning (TLRL) algorithm toconsider taskof oading forresource-constrained mobiledevices. Asopposed toexisting literature, the utilization rate of the physical machine and the delay for of oaded tasks are taken into account simultaneously by introducing a weighted reward. The high dimensionality of the state space and action space might affect the speed of convergence. Therefore, a novel reinforcement learning algorithm with a two-layered structure is presented to address this problem. First, k clusters of the physical machines are generated based on the k-nearest neighbors algorithm (k-NN). The rst layer of TLRL is implemented by a deep reinforcement learning to determine the cluster to be assigned for the of oaded tasks. On this basis, the second layer intends to further specify a physical machine for task execution. Finally, simulation examples are carried out to verify that the proposed TLRL algorithm is able to speed up the optimal policy learning and can deal with the tradeoff between physical machine utilization rate and delay.
Mingxing Jiang, Min Hu, Xiaohua Wang, Fuji Ren and Haowen Wang : Dual-Modal Emotion Recognition Based on Facial Expression and Body Posture in Video Sequences, Laser & Optoelectronics Progress, Vol.55, 161-168, 2018.
Zhao Han, Duoqian Miao and Fuji Ren : Rough Set Based Knowledge Predicate Analysis of Chinese Knowledge Based Question Answering, COMPUTER SCIENCE, Vol.45, No.6, 183-186, 2018.
(Summary)
In knowledge based question answering system, the performance of knowledge predicate analysis can affect the overall match result of knowledge triple The knowledge predicate analysis of Chinese short questionis difficult because of the uncertainty of Chinese knowledge predicate representation Based on the rough set theory, a new definition of knowledge predicate analysis of knowledge based questions answering was given, and a new method was proposed to analyze the knowledge predicate of question It can reduce the words which are weakly related with the knowledge predicate, and then the words which are more related with knowledge predicate representation will be used to match the knowledge triples to improve the overall performance of system The experiment results verify the validity of the method
Tian Chen, Jiawei Wang, Xin An and Fuji Ren : Parallel test scheduling optimization method for three-dimensional chip with multi-core and multi-layer, Journal of Computer Applications, Vol.38, No.6, 1795-1800, 2018.
(Summary)
In order to solve the problem of high cost of chip testing in the process of Three-Dimensional ( 3D) chip manufacturing a new scheduling method based on Time Division Multiplexing ( TDM) was proposed to optimize the testing resources between layers layer and core cooperatively Firstly the shift registers were arranged on each layer of 3D chip and the testing frequency was divided properly between the layers and cores of the same layer under the control of shift register group on input data so that the cores in different locations could be tested in parallel Secondly greedy algorithm was used to optimize the allocation of registers for reducing the free test cycles of core parallel test Finally Discrete Binary Particle Swarm Optimization ( DBPSO) algorithm was used to find out the best 3D stack layout so that the transmission potential of the Through Silicon Via ( TSV) could be adequately used to improve the parallel testing efficiency and reduce the testing time The experimental results show that under the power constraints the utilization rate of the optimized whole Test Access Mechanism ( TAM) is increased by an average of 16 28% and the testing time of the optimized 3D stack is reduced by an average of 13 98% The proposed method can decrease the time and reduce the cost of testing
Kazuyuki Matsumoto, Kyosuke Akita, Fuji Ren, Minoru Yoshida and Kenji Kita : Intimacy Estimation of the Characters in Drama Scenario, Journal of Japan Society for Fuzzy Theory and Intelligent Informatics, Vol.30, No.3, 591-604, 2018.
(Summary)
Recently, a portable digital device equipped with voice guidance has been widely used, increasing the demand for the usabilityconscious dialogue system. One of the problems with the existing dialogue system is its immature application to non-task-orienteddialogue. Non-task-oriented dialogue requires some scheme that enables smooth and flexible conversations with a user. For example,it would be possible to provide topics related to a person who is familiar with a user or avoid providing topics related to a personwho is not a good friend with the user, by considering relationship with others in real life of the user. In this paper, we focus on thedialogue made by the two characters in a drama scenario, and tried to express their relationship with a scale of intimacy degree.There will be such various elements related to the intimacy degree as the frequency of response to the utterance and the attitude ofa speaker during the dialogue. We focus on the emotional state of the speaker during the utterance and try to realize an intimacyestimation with high accuracy. As the evaluation result, we achieved high accuracy in intimacy estimation than the existing methodbased on speech role.
(Keyword)
intimacy degree / scenario dialogue / emotional state
Zhao Han, Duoqian Miao, Fuji Ren and Hongyun Zhang : Rough Set Knowledge Discovery Based Open Domain Chinese Question Answering Retrieval, Journal of Computer Research and Development, Vol.55, No.5, 958-967, 2018.
Liangfeng Xu, Yonghai Liu, Min Hu, Xiaohua Wang and Fuji Ren : Speech emotion recognition based on improved complete local binary pattern of spectrogram images, Journal of Electronic Measurement and Instrument, Vol.32, No.5, 25-32, 2018.
(Summary)
A novel speech emotion recognition is proposed in the paper in whichthe improved complete local binary pattern and the power exponential class weighted fusion are used Firstly the original speech sample is transformedinto a spectrum diagram then the 5scale 8-directional Log-Gabor filter is used to deal with the language spectrum in order to enlarge the detail information of the image Then the block histogram features of the uniform complete local binary pattern sign feature and improved complete local binary patternmagnitude feature are extracted and cascaded as a new fusion feature called improved complete local binary patternsign magnitude Finally based on support vector machine these three features are weighted by the decision level in order to achieve the speech emotion recognition The experiment results indicate that theimproved complete local binary patternmagnitude feature and the fusion feature can reduce the feature dimension and improve the effect of system identification The exponential weighted fusion method expands the gap between classifiers and gives a larger weight to the classifier with better classification performance which can finally enhance the performance of speech emotion recognition effectively Compared with other algorithms the experiment results also show the effectiveness of the proposed algorithm
Fuji Ren and Ning Liu : Emotion computing using Word Mover's Distance features based on Ren_CECps, PLoS ONE, Vol.13, No.4, 1-17, 2018.
(Summary)
In this paper, we propose an emotion separated method(SeTF·IDF) to assign the emotion labels of sentences with different values, which has a better visual effect compared with the values represented by TF·IDF in the visualization of a multi-label Chinese emotional corpus Ren_CECps. Inspired by the enormous improvement of the visualization map propelled by the changed distances among the sentences, we being the first group utilizes the Word Mover's Distance(WMD) algorithm as a way of feature representation in Chinese text emotion classification. Our experiments show that both in 80% for training, 20% for testing and 50% for training, 50% for testing experiments of Ren_CECps, WMD features get the best f1 scores and have a greater increase compared with the same dimension feature vectors obtained by dimension reduction TF·IDF method. Compared experiments in English corpus also show the efficiency of WMD features in the cross-language field.
Xiao Sun, Chen Zhang and Fuji Ren : User Emotion Modeling and Anomaly Detection Based on Social Media, Journal oc Chinese Information Processing, Vol.32, No.4, 120-129, 2018.
(Summary)
For abnormal emotional detection among micro-blog users, this paper proposes ananomaly detection method based on the joint probability density of multivariate Gaussian model and power-law distribution. In the xperiments, the anomaly detection accuracy is 83. 49% in terms of individual user, and 87. 84% in terms of month. Statistics reveals that individual users' neutral, happy and sad emotions fall into the normal distribution, but the amazed and angry emotions are not. Emotions of micro-blogs released by groups confirm to the power law distribution, but not those by the individual.
100.
Yiming Tang, Gangyong Feng, Fuji Ren, Xianghui Hu and Youcheng Zhang : Fuzzy clustering validity index facing data set with complexity structure, Journal of Electronic Measurement and Instrument, Vol.32, No.4, 119-127, 2018.
(Summary)
How to effectively determine the clustering number is one of the historical challenges in the clustering field A typical algorithm for clustering is the fuzzy c-means algorithm Nowadays it is difficult for clustering validity index to make accurate judgment for complex data structure and the huge disparity of cluster size Aiming at the problem a new clustering validity index VGSDC( index for geometry structure and different clusters) is proposed for the geometry structure of data set and clusters with different size The compactness strategy is obtained by square error sum within the classes as well as weights of membership degree and then the separation computing method is gotten by minimum distance between cluster centers together with distance sum from centers to average center And finally the new validity index VGSDC is obtained Furthermore the best clustering number can be automatically obtained by the number of categories corresponding to the extreme value of VGSDC Through the experiment in 6 data structure it is found that the proposed VGSDC gets the better performance comparing with 11 kinds of clustering validity indexes VGSDC can not only handle multiple types of data sets and also provide full consideration to the structure characteristics and complexity of the data sets As a result it can be applied to large data sets with huge differences of clustering center distance
Yiming Tang, Youcheng Zhang, Fuji Ren, Xianghua Hu, Xiaocheng Song and Gangyong Feng : FMT symmetric I* method of fuzzy reasoning, Journal of Nanjing University (Natural Science), Vol.54, No.4, 706-713, 2018.
102.
Jun Liu, Chengqiang Zhu, Xi Wu, Wei Wang and Fuji Ren : Yield Optimization Technique for Three Dimensional Memory Based on Redundancy Sharing Among Adjacent Layers, ACTA Electronica Sinica, Vol.46, No.3, 629-635, 2018.
Jing Zhang, Degen Huang, Kaiyu Huang, Zhuang Liu and Fuji Ren : Corpus Expansion for Neural CWS on Microblog-Oriented Data with λ-Active Learning Approach, IEICE Transactions on Information and Systems, Vol.E101-D, No.3, 778-785, 2018.
(Summary)
Microblog data contains rich information of real-world events with great commercial values, so microblog-oriented natural language processing (NLP) tasks have grabbed considerable attention of researchers. However, the performance of microblog-oriented Chinese Word Segmentation (CWS) based on deep neural networks (DNNs) is still not satisfying. One critical reason is that the existing microblog-oriented training corpus is inadequate to train e ective weight matrices for DNNs. In this paper, we propose a novel active learning method to extend the scale of the training corpus for DNNs. However, due to a large amount of partiallyoverlappedsentencesinthemicroblogs,itisdi culttoselectsamples with high annotation values from raw microblogs during the active learning procedure. To select samples with higher annotation values, parameter λ is introduced to control the number of repeatedly selected samples. Meanwhile, various strategies are adopted to measure the overall annotation values of a sample during the active learning procedure. Experiments on the benchmark datasets of NLPCC 2015 show that our λ-active learning method outperforms the baseline system and the state-of-the-art method. Besides, the results also demonstrate that the performances of the DNNs trained on the extended corpus are signi cantly improved.
Xiaohua Wang, Chen Xia, Min Hu and Fuji Ren : Facial Expression Recognition Based on the Fusion of Spatio-temporal Features in Video Sequences, Journal of Electronics and Information Technology, Vol.40, No.3, 626-632, 2018.
(Summary)
For facial expression recognition based on video sequences, the changing information of facial regions along the time axis can be described by dynamic descriptors more effectively than static descriptors. This paper proposes an expression recognition method based on the dynamic texture and motion information, learning from the principle of Local Binary Pattern on Three Orthogonal Planes (LBP-TOP), Spatio-Temporal Weber Local Descriptor (STWLD) is proposed to describe the dynamic texture feature information of the facial expression sequence. Moreover, using Block-based Histogram of Optical Flow features (BHOF), the motion information can be described. Through the combination of the dynamic texture and motion information, and finally SVM is applied to complete the expression classification. The results of the cross experiments on the CK + and MMI expression database show that the method achieves better performance than methods using the single descriptors. The comparison experiments with other related methods also prove the superiority of the method.
Xiao Sun, Chen Zhang, Guoqiang Li, Daniel Sun, Fuji Ren, Albert Zomaya and Rajiv Ranjan : Detecting Users' Anomalous Emotion Using Social Media for Business Intelligence, Journal of Computational Science, Vol.25, 193-200, 2018.
(Summary)
Anomaly detection in sentiment analysis refers to detecting users' abnormal opinions, sentiment patterns or special temporal aspects of such patterns. Users' emotional state extracted from social media contains business information and business value for decision making. Social media platforms, such as Sina Weibo or Twitter, provide a vast source of information, which include user feedbacks, opinions and information on most issues. Many organizations also leverage social media platforms to publish information about events, products, services, policies and other topics frequently, analyzing social media data to identify abnormal events and make decisions in a timely manner is a beneficial topic. This paper adopts the multivariate Gauss distribution with the power-law distribution to model and analyze the users' emotion of micro-blogs and detect abnormal emotion state. With the measure of joint probability density value and the validation of the corpus, anomaly detection accuracy of individual user is 83.49% and of different month is 87.84% by this method. Through the distribution test, the results show that individual users' neutral, happy and sad emotions obey the normal distribution, but the surprised and angry emotions do not. Besides, emotions of micro-blogs released by groups obey power-law distribution, but the individual emotions do not. This paper proposes a quantitative method for abnormal emotion detection on social media, which automatically captures the correlation between different features of the emotions, and saves a certain amount of time by batch calculation of the joint probability density of data sets. The method can help the businesses and government organizations to make decisions according to the user's affective disposition, intervene early or adopt proper strategies if needed.
(Keyword)
Business intelligence / Sentiment analysis / Anomaly detection / Multivariate Gaussian distribution / Decision making
XIN KANG, Fuji Ren and Yunong Wu : Exploring latent semantic information for textual emotion recognition in Blog articles, IEEE/CAA Journal of Automatica Sinica, Vol.5, No.1, 204-216, 2018.
(Keyword)
emotion recognition / natural language understanding / emotion-topic model / Bayesian inference / multi-label classification
Taihao Li, Tuya Naren, Jianshe Zhou, Fuji Ren and Shupeng Liu : An Improved K Means Algorithm Based on Initial Clustering Center Optimization, ZTE Communications, Vol.15, No.S2, 43-46, 2017.
(Summary)
The K means algorithm is widely known for its simplicity and fastness in text clustering. However, the selection of the initial clus tering center with the traditional K means algorithm is some random, and therefore, the fluctuations and instability of the cluster ing results are strongly affected by the initial clustering center. This paper proposed an algorithm to select the initial clustering center to eliminate the uncertainty of central point selection. The experiment results show that the improved K means clustering algorithm is superior to the traditional algorithm.
Fuji Ren and Kazuyuki Matsumoto : Emotion Analysis on Social Big Data, ZTE Communications, Vol.15, No.S2, 30-37, 2017.
(Summary)
In this paper, we describe a method of emotion analysis on social big data. Social big data means text data that is emerging on In ternet social networking services.We collect multilingual web corpora and annotated emotion tags to these corpora for the purpose of emotion analysis. Because these data are constructed by manual annotation, their quality is high but their quantity is low. If we create an emotion analysis model based on this corpus with high quality and use the model for the analysis of social big data, we might be able to statistically analyze emotional sensesand behavior of the people in Internet communications, which we could not know before. In this paper, we create an emotion analysis model that integrate the high quality emotion corpus and the automatic constructed corpus that we created in our past studies, and then analyze a large scale corpus consisting of Twitter tweets based on the model. As the result of time series analysis on the large scale corpus and the result of model evaluation, we show the effective ness of our proposed method.
Minjia Li, Lun Xie, Zhiliang Wang and Fuji Ren : Emotion and Cognitive Reappraisal Based on GSR Wearable Sensor, ZTE Communications, Vol.15, No.S2, 18-22, 2017.
(Summary)
Various wearable equipment enables us to measure people behavior by physiological signals. In our research, we present one gal vanic skin reaction (GSR) wearable sensor which can analyze human emotions based on cognition reappraisal. First, We research the factors of emotional state transition in Arousal Valence Stance(AVS) emotional space. Second, the influence of the cognition on emotional state transition is considered, and the reappraisal factor based on Gross regulation theory is established to correct the effectiveness from cognitive reappraisal ability to emotional state transition. Third, based on the previous work, we establish a GSR emotion sensing system for predicting emotional state transition and considering the correlation between GSR signals and emotions. Finally, an overall wearable sensor layout is built. In the experiment part, we invited 30 college students to wear our GSR sensors and watch 14 kinds of affective videos. We recorded their GSR signals while asking them to record their emotional states synchronously. The experiment results show different reappraisal factors can predict subjects'emotional state transition well and indirectly confirm the feasibility of the Gross regulation theory.
Fuji Ren, Yanqiu Li and Min Hu : Multi-classifier ensemble based on dynamic weights, Multimedia Tools and Applications, Vol.9, No.2, 1-15, 2017.
(Summary)
In this study, a novel multi-classifier ensemble method based on dynamic weights is proposed to reduce the interference of unreliable decision information and improve the accuracy of fusion decision. The algorithm defines decision credibility to describe the realtime importance of the classifier to the current target, combines this credibility with the reliability calculated by the classifier on the training data set and dynamically assigns the fusion weight to the classifier. Compared with other methods, the contribution of different classifiers to fusion decision in acquiring weights is fully evaluated in consideration of the capability of the classifier to not only identify different sample regions but also output decision information when identifying specific targets. Experimental results on public face databases show that the proposed method can obtain higher classification accuracy than that of single classifier and some popular fusion algorithms. The feasibility and effectiveness of the proposed method are verified.
Kazuyuki Matsumoto, Fuji Ren, Minoru Yoshida and Kenji Kita : Review Score Estimation Based on Transfer Learning of Different Media Review Data, International Journal of Advanced Intelligence (IJAI), Vol.9, No.4, 541-555, 2017.
(Summary)
We propose a model to classify reviews based on review data from different media sources. Recently, research has been actively conducted on transfer learning between different domains with various kinds of big data as the target. The fact that evaluation expressions often vary in different domains presents a barrier to reputation analysis. Users commonly use various linguistic expressions to refer to creative works, depending on the specific media form.For example, the terms or expressions used in anime to describe creative works within that medium are different from the expressions used in comics, or games or movies. These differences can be considered as features of each individual medium. We should expect, then, that there would be differences in evaluation expressions among the various media, as well. We analyze the effects of such differences on classification accuracy by conducting transfer learning between review data from different media and demonstrate compatibility between the original (pre-transfer) and target (post-transfer) media by constructing a review classification model. As a result of our evaluation experiments, we are able to more accurately estimate review scores without using SO-Scores for training review fragments based on Long Short-Term Memory (LSTM) rather than using a method based on SO-Scores.
(Keyword)
review classification / transfer learning / Long Short-Term Memory / different media
Xiao Sun, Man LV, Changqin Quan, Fang Tian, Fuji Ren and Kunxia Wang : Fine-Grained Emotion Elements Extraction and Tendency Judgment Based on Mixed Model, Information Engineering Express, Vol.3, No.4, 21-32, 2017.
(Summary)
Nowadays, with the development of internet technology and electronic commerce, the Web storages huge number of product reviews comment by customers. Product reviews tend to be more objective in reflecting the real situation of the product, more and more customers post product reviews at merchant websites in order to make an informed choice. However, a large number of reviews made it difficult to track the comments and suggestions that customers made. In this paper, a fine-grained emotional element detection and emotional tendency judgment method based on conditional random fields (CRFs) and support vector machine (SVM) was proposed. This model introduces semantics and word meaning in CRF model to improve the robustness. In SVM model, deep semantic information imported based on neural network to improve the traditional bag of word model. Experimental results show that the proposed model with deep features efficiently improved the F-Measure.
(Keyword)
俗語 / トピック分析 / 時系列分析
114.
Yu Gu, Jinhai Zhan, Yusheng Ji, Jie Li, Fuji Ren and Shangbin Gao : MoSense: An RF-Based Motion Detection System via Off-the-Shelf WiFi Devices, IEEE Internet of Things Journal, Vol.4, No.6, 2326-2341, 2017.
(Summary)
Motion is a critical indicator of human presence and activities. Recent developments in the field of indoor motion detection have revealed their potentials in enhancing our living experiences through applications like intrusion detection and sleep monitoring. However, existing solutions still face several critical downsides such as the availability (specialized hardware), reliability (illumination and line-of-sight constraints), and privacy issues (being watched). To overcome such shortages, a radio frequency (RF) based device-free motion detection system (MoSense) is designed via leveraging the attenuation of ubiquitous WiFi signals induced by motions to deliver a reliable and transparent detection service in realtime. The design and implementation of MoSense face two challenges: 1) characterizing stationary states and 2) the noisy subcarriers. For the first challenge, a silence analysis model is proposed to characterize stationary states for distinguishing motions. For the second challenge, we design a distance-based mechanism to select certain subcarriers that better capture the impact of motions from the noisy channel through measuring the similarity between subcarriers. A prototype of MoSense is realized and evaluated in real environments. By comparing MoSense with other two state-of-the-art systems, i.e., FIMD and FRID, we have shown that MoSense is superior in terms of precision, false negative rate and computational complexity. Considering that MoSense is compatible with existing WiFi infrastructure, it constitutes a low-cost yet promising solution for motion detection.
Yunong Wu, Kenji Kita, Kazuyuki Matsumoto, XIN KANG and Fuji Ren : Constructing A Short Text Conversation system Based on the Relations Between Posts and Comments, International Journal of Advanced Intelligence (IJAI), Vol.9, No.3, 369-380, 2017.
116.
Xiao Sun, Xiaoqi Peng, Min Hu and Fuji Ren : Extended Multi-modality Features and Deep Learning Based Microblog Short Text Sentiment Analysis, Journal of Electronics and Information Technology, Vol.39, No.9, 2048-2055, 2017.
(Summary)
This paper presents a Deep Belief Nets (DBN) model and a multi-modality feature extraction method to extend features , dimensionalities of short text for Chinese microblogging sentiment classification. Besides traditional features sets for document classification, comments for certain posts are also extracted as part of the microblogging features according to the relationship between commenters and posters through constructing microblogging social network as input information. Multi-modality features are combined and adopted as the input vector for DBN. A DBN model, which is stacked with several layers of Restricted Boltzmann Machine (RBM), is implemented to initialize the structure of neural network. The RBM layers can take probability distribution samples of input data to learn hidden syntactic structures for better feature representation. A Classification RBM (ClassRBM) layer, which is stacked on top of the former RBM layers, is adapted to achieve the final sentiment classification. The results demonstrate that, with proper structure and parameter, the performance of the proposed deep learning method on sentiment classification is better than the state of the art surface learning models such as SVM or NB, which proves that DBN is suitable for short-length document classification with the proposed feature dimensionality extension method.
Heng Zhang, Shupeng Liu, Xiangfei Yang, Na Chen, Fufei Pang, Zhenyi Chen, Tingyun Wang, Jianshe Zhou, Fuji Ren, Xiaoyin Xu and Taihao Li : LED Phototherapy With Gelatin Sponge Promotes Wound Healing in Mice, Photochemistry and Photobiology, 2017.
(Summary)
Tiny but highly efficient, a light emitting diode (LED) can power a therapy device, such as a phototherapy device, and, at the same time, decrease the device's size requirements. In this study, a LED phototherapy device was designed to investigate the possible impact on wound healing using a mouse model and a cell line exposed to red and blue light. To enhance wound phototherapy, a gelatin sponge was fabricated. Results showed that the red and blue lights promoted cell growth and wound healing, while the blue light with a gelatin sponge protected the wound from infection in the early stages of wound healing. The LED phototherapy device combined with the gelatin sponge, therefore, has potential significance in clinical application for wound healing.
Jun Liu, Jin Li, Wei Wang and Fuji Ren : A Stacking Scheme Using Adjacent Redundancy Across Dies for 3D-stacked Memory, Microelectronics & Computer, Vol.34, No.7, 1-6, 2017.
(Keyword)
俗語 / トピック分析 / 時系列分析
119.
Changqin Quan, Bin Zhang, Xiao Sun and Fuji Ren : A combined cepstral distance method for emotional speech recognition, International Journal of Advanced Robotic Systems, Vol.14, No.4, 1-9, 2017.
(Summary)
Affective computing is not only the direction of reform in artificial intelligence but also exemplification of the advanced intelligent machines. Emotion is the biggest difference between human and machine. If the machine behaves with emotion, then the machine will be accepted by more people. Voice is the most natural and can be easily understood and accepted manner in daily communication. The recognition of emotional voice is an important field of artificial intelligence. However, in recognition of emotions, there often exists the phenomenon that two emotions are particularly vulnerable to confusion. This article presents a combined cepstral distance method in two-group multi-class emotion classification for emotional speech recognition. Cepstral distance combined with speech energy is well used as speech signal endpoint detection in speech recognition. In this work, the use of cepstral distance aims to measure the similarity between frames in emotional signals and in neutral signals. These features are input for directed acyclic graph support vector machine classification. Finally, a two-group classification strategy is adopted to solve confusion in multi-emotion recognition. In the experiments, Chinese mandarin emotion database is used and a large training set (1134 + 378 utterances) ensures a powerful modelling capability for predicting emotion. The experimental results show that cepstral distance increases the recognition rate of emotion sad and can balance the recognition results with eliminating the over fitting. And for the German corpus Berlin emotional speech database, the recognition rate between sad and boring, which are very difficult to distinguish, is up to 95.45%.
Fuji Ren and Lei Wang : Sentiment analysis of text based on three-way decisions, Journal of Intelligent and Fuzzy Systems, Vol.33, No.1, 245-254, 2017.
(Summary)
In recent years, affective computing has received much attention in the area of natural language processing and arti cial intelligence. Sentiment orientation recognition of text is one of important parts in affective computing. A methodisproposedtorecognizethemulti-labelsentimentorientationsofChinesetextbasedonthree-waydecisions.Firstly, sentiment orientation and intensity of sentiment words from texts are identi ed by sentiment lexicons, Tongyi Cilin and HowNet.Subsequently,sentimentorientationoftextaredividedintothreedomains,includingpositive,negativeandboundary domain,accordingtotheirsentimentintensityandtheappropriatedecision-makingthresholds.Lastly,sentimentorientations of texts in the boundary domain are distinguished pursuant to sentimental characteristics of sentences in texts. The results of experiments show that the method of multi-label sentiment analysis of Chinese text, based on three-way decisions, is effective for identifying sentiment orientations of texts.
Hiroki Urakami, Shun Nishide and Fuji Ren : Towards a Developmental Human-Robot Interaction System Using Robot Facial Expressions From Human Feedback, International Journal of Advanced Intelligence (IJAI), Vol.9, No.2, 127-136, 2017.
(Summary)
Creation of a developmental structure for a human-robot interaction model is essential for practical human robot interaction. We speci cally focus on creating a developmental system of the robot's facial expression based on spoken dialogue of the robot. A humanoid robot Actroid is used to create smooth facial expressions. In the proposed method, we rst create a xed dialogue system with a prede ned facial expression related to each of the robot's dialogue. During communication with a human subject, a feedback (judging if the robot's facial expression was natural or not) is given by the human after the robot's utterance. The facial expression related with the dialogue is changed randomly to a di erent facial expression if the human's feedback denote the robot's expression as unnatural. We constructed the feedback using two types of methods: button pressing and auditory. The actual experiments show that the robot is capable of acquiring natural facial expressions after several communication trials with the human subject.
(Keyword)
Human robot interaction / Developmental model / Humanoid robot
122.
Kazuyuki Matsumoto, Satoshi Tanaka, Minoru Yoshida, Kenji Kita and Fuji Ren : Ego-state Estimation from Short Texts Based on Sentence Distributed Representation, International Journal of Advanced Intelligence (IJAI), Vol.9, No.2, 145-161, 2017.
(Summary)
Human personality multilaterally consists of complex elements. Egogram is a method to classify personalities into the patterns according to the combinations of the five levels of ego-states. With recent development of Social Networking Service (SNS), more researches have been trying to judge personality from the statements on SNS. However, there are several problems in the personality judgment based on the superficial information of the statements. Concretely, personality is not always reflected on every statement and the tendency of the statements with influence of personality changes with the times. It is also important to collect sufficient amount of statement data including the results of personality judgment. In this paper, to realize automatic egogram judgement, we focused on the short texts on the SNS, especially microblogs, and represented the comments on Twitter by distributed representation (sentence vector) in the pre-training. Then, we tried to create a model to estimate ego-state level of each user by using a deep neural network. The experimental result showed that our proposed method estimated ego-state with higher accuracy than the baseline method based on bag of words. To investigate the change of personality according to time, we also analyzed how the match rates of the estimation results changed before/after the egogram judgment. Moreover, we confirmed that the personality pattern classification was improved by adding a feature expressing the formal degree of the sentence.
(Keyword)
Egogram / Personality Estimation / Twitter / Social Networking Service / Distributed Representation
Xun Feng, Hongjun Ni and Fuji Ren : Voice conversion based on RBF neural network, International Journal of Advanced Intelligence (IJAI), Vol.9, No.2, 163-173, 2017.
(Summary)
Recently, voice conversion has becoming the research hotspot, because of its widely application areas. However, the voice conversion technology is still immature. By the researching of existing voice conversion models, the voice conversion system based on the RBF neutral network was designed, and the system simulation was implemented. During conversion, the unvoiced speech was excluded and the voiced speech was reserved. The LPC was the extracted from the source and target speech, then convert the LPC to LSP. The LPS was trained by RBF neural network after time-aligned. Obtained mapping function was used to convert the source LSP to target LSP, and synthesis the speech. Finally, the converted speech evaluated by ABX and MOS to test the tendency and quality of the speech.
(Keyword)
Voice conversion / RBF / Neutral network
124.
Min Hu, Yaona Zheng, Xiaoyin Huang and Fuji Ren : FacialExpressionRecognitionBasedonMonogenicMulti-featureandFused SparseRepresentation, International Journal of Advanced Intelligence (IJAI), Vol.9, No.1, 111-126, 2017.
(Summary)
In the eld of facial expression recognition, in order to make up for insuf cient information of monogenic magnitude and orientation which is extracted by monogenic binary pattern, and enhance robustnessofsparseclassi erforfacialexpressionclassi cation,afacialexpressionrecognitionmethodbased on monogenic multi-feature and fused sparse representation is proposed in this paper. First of all, the preprocessed images of facial expression are ltered by the monogenic signal with the purpose of acquiringtheinformationofmonogenicmagnitude,orientation,andphase.Secondly,wefusemagnitude, orientation and phase information with the method of monogenic binary pattern, monogenic oriented gradient, and enhanced monogenic phase respectively, to form different expression features and construct the corresponding classi ers. Finally facial expression recognition is completed by using the regularized least-square theory to optimize the weight of double classi er. Experiments are performed on JAFFE and Cohn-Kanade facial expression databases and the average rates reach up to 97.30% and 99.33% respectively. The results of experiments show that the proposed method effectively improves the recognition ef ciency of facial expression.
Shun Nishide, Hidenobu Shibasaki, XIN KANG and Fuji Ren : Generation of Humanlike Facial Expression for Natural Human-Robot Interaction System, International Journal of Advanced Intelligence (IJAI), Vol.9, No.1, 77-65, 2017.
(Summary)
Recognition and expression of emotions are indispensible functions for robots intended to play a role in the human society. Among various modalities, facial expression is said to be one of the most important factors for represent emotions. In this paper, we present our work on creating humanlike facial expressions for the humanoid robot Actroid, and adapting it to actual human-robot interaction scenario. Facial expressions were created manually using the software Wten source code, by adjusting parameters of facial features. The created expressions were implemented in a xed dialogue between a human and robot. Experiments were conducted with human subjects talking with the robot. Evaluations were done using questionaires on naturalness and e ectiveness of facial expressions during the dialogue.
(Keyword)
Human robot interaction / Facial expression / Humanoid robot
126.
Jiawen Deng, Mingyu Huang and Fuji Ren : Text Classification Based on Word Co-occurrence with Background knowledge, International Journal of Advanced Intelligence (IJAI), Vol.9, No.1, 29-42, 2017.
(Summary)
This article proposes a data treatment strategy to generate a new text representation model with discriminative feature for the sake of text classification. This feature is determined by word co-occurrence information, and is extracted by measuring the cosine similarity between word co-occurrence vectors of testing documents and background knowledge in Chinese language. The background knowledge is composed by the word co-occurrence matrix of all words in background corpus based on top n keywords in each category which selected by TFIDF value. This strategy is to get discriminative features to reduce the ambiguity and noise inherent of the traditional representation model. In the experiments, using a linear SVM classifier, we investigate the effects of the proposed method with two Chinese classification corpora, and two background corpora are used as background knowledge. Results show that, compared with conventional method, the proposed strategy performs better in the values of precision, recall and F1 score.
(Keyword)
Text classification / text representation / word co-occurrence / background knowledge
127.
XIN KANG, Yunong Wu and Fuji Ren : Disambiguating Users' Temporal Intent in Search Queries with Deep Neural Networks, International Journal of Advanced Intelligence (IJAI), Vol.9, No.1, 11-28, 2017.
128.
Xiao Sun, Fuji Ren and Jiaqi Ye : Trends Detection of Flu based on Essemble Models with Emotional Factors from Social Networks, IEEJ Transactions on Electrical and Electronic Engineering (TEEE), Vol.12, No.3, 388-396, 2017.
(Summary)
In uenza is an acute respiratory illness and widespread activity that occurs every year. Detection and prevention of in uenza in its earliest stage would reduce the spread range of the illness. Sina microblog is a popular microblogging service in China, which can be treated as perfect reference sources for u detection because of its real-time character. A large number of active users post about their daily life continually. In this paper, we investigate the real-time u detection problem and propose a u detection model with emotion factors and semantic information. First, we extract u-related microblog posts automatically in real time by adopting support vector machine (SVM) lter and semantic features. We use association rule mining to extract strongly associated features as additional features for posts to overcome the limitation of 140 words, including sentiment information, which can help us to classify the posts without u-related features. Then, the conditional random eld model is revised and applied to detect the transition time of u so that we can nd out which place is more likely to have in uenza outbreak and when it is more likely to have in uenza outbreak in a city or province in China. Experimental results on detecting u situation during certain times in some locations show the robustness and effectiveness of the proposed model, which might help health authorities in predicting u outbreak ahead and take timely control action and response.
Yiming Tang and Fuji Ren : Fuzzy Systems Based on Universal Triple I Method and Their Response Functions, International Journal of Information Technology & Decision Making, Vol.16, No.2, 443-471, 2017.
(Summary)
The fuzzy systems based on the universal triple I method are investigated, and then theirresponse functions are analyzed. First, the conclusions show that 100 fuzzy systems via theuniversal triple I method are approximately interpolation functions, which can be used inpractical systems, and that 90 ones are approximately ¯tted functions, which may be usable.Second, as its special cases, the Compositional Rule of Inference (CRI) method and the triple Imethod are discussed, with the results that 19 fuzzy systems via the CRI method and 2 ones viathe triple I method are practicable. Therefore, the universal triple I method has larger e®ectivechoosing space, which can obtain more usable fuzzy systems than the others. Lastly, it is foundthat the ¯rst implication and second implication, respectively, embody the function of rule baseand reasoning mechanism, further demonstrating the reasonability of the universal triple Imethod.
Xiaohua Wang, Dengyong Hou, Min Hu and Fuji Ren : Dual-modality emotion recognition based on composite spatio-temporal features, Journal of Image and Graphics, Vol.22, No.1, 39-48, 2017.
(Summary)
Objective In view of existing algorithms, volume local binary pattern is applied to the feature extraction of video frames. However, problems such as large feature dimension, weak robustness to illumination, and noise exist. This study proposes a new feature description algorithm, which is temporal-spatial local ternary pattern moment. This algorithm introduces three value patterns, and it is extended to the temporal-spatial series to describe the variety of pixel values among adjacent frames. The value of texture feature is represented by the energy values of the three value model matrixes, which are calculated according to the gray-level co-occurrence matrix. Considering that the temporal-spatial local ternary pattern moment only describes the texture feature, it lacks the expression of image edge and direction information. Therefore, it cannot fully describe the characteristics of emotional videos. The feature of 3D histograms of oriented gradients is further fused to enhance the description of the emotion feature. Composite spatio-temporal features are obtained by combining two different features. Method First, the emotional videos are preprocessed, and five frame images are obtained by K mean clustering, which are used as the expression and body posture emotion sequences. Second, TSLTPM and 3DHOG features are extracted from the expression and gesture emotion sequences, and the minimum Euclidean distance of the feature between the test sequence and labeled emotion training set is calculated. The calculated value is used as independent evidence to construct the basic probability assignment function. Finally, according to the rules of D-S evidence theory, the expression recognition result is obtained by fused BPA. Result Experimental results on the bimodal expression and body posture emotion database show that complex spatio-temporal features exhibit good recognition performance. The average recognition rates of 83.06% and 94.78% are obtained in the single model identification of facial expressions and gestures, respectively, compared with other algorithms. The average recognition rate of the single-expression model is 9.27%, 12.89%, 1.87%, and 1.13% higher than those of VLBP, LBP-TOP, TSLTPM, and 3DHOG, respectively. The average recognition rate of the single-gesture model is 24.61%, 27.55%, 1.18%, and 0.98% higher than those of VLBP, LBP-TOP, TSLTPM, and 3DHOG, respectively. The average recognition rate after the fusion of these two models is 96.86%, which is higher than the rate obtained by a single model. This result confirms the effectiveness of emotion recognition under the fusion of expression and gesture. Conclusion The TSLTPM feature proposed in our paper extends the VLBP, which is effective in describing the local features of video images, into the temporal-spatial local ternary pattern. The proposed feature has low dimensionality, and it can enhance the robustness to illumination and noise. The composite spatio-temporal features fused with 3DHOG and TSLTPM can fully describe the effective information of emotional videos, and it enhances the classification performance of such videos. The effectiveness of the proposed algorithm in comparison with other typical feature extraction algorithms is also demonstrated. The proposed algorithm is proven suitable for identifying the emotion of static background videos, and the superiority of the fusion method in this study is verified.
Min Hu, Zixi Yu, Xiaohua Wang, Fuji Ren and Lei He : G-LBP and Variance Cross Projection Function for Face Recognition, Journal of Graphics, Vol.38, No.1, 82-89, 2017.
(Summary)
In order to enhance robustness of traditional Gabor features towards illumination, expression and pose variance and overcome its high dimension problem, the paper proposes a face recognition method based on Gabor, local binary patter and variance projection entropy improved algorithm. First, the multi direction multi-scale fusion Gabor image is coded with LBP, and the coded image fused and the histograms of image block calculated. Second, a local projection entropy feature extraction is adopted for face images with anti-geometric distortion variance projection entropy and cross variance projection entropy operator. Finally, the face recognition is completed by using BP neutral network to fuse and make decision weightily. The G-LBP feature extraction reduces the redundancy of data greatly, and maintains the integrity of the effective information. Variance projection of entropy and cross entropy improves the richness of the feature. The weighted fusion in decision-making layer plays an important role of integration between the classifiers and improves the recognition rate of face recognition. Compared with other literature algorithms, experiment results verify the effectiveness and superiority of the proposed algorithm.
Xiao Sun, Chongyuan Sun, Changqin Quan, Fuji Ren, Fang Tian and Kunxia Wang : Fine-Grained Emotion Analysis Based on Mixed Model for Product Review, International Journal of Networked and Distributed Computing, Vol.5, No.1, 1-11, 2017.
(Summary)
Nowadays, with the rapid development of B2C e-commerce and the popularity of online shopping, the Web storages huge number of product reviews comment by customers. A large number of reviews made it difficult for manufacturers or potential customers to track the comments and suggestions that customers made. This paper presents a method for extracting emotional elements containing emotional objects and emotional words and their tendencies from product reviews based on mixed model. First we constructed conditional random fields to extract emotional elements, lead-in semantic and word meaning as features to improve the robustness of feature template and used rules for hierarchical filtering errors. Then we constructed support vector machine to classify the emotional tendency of the fine-grained elements to achieve key information from product reviews. Deep semantic information imported based on neural network to improve the traditional bag of word model. Experimental results show that the proposed model with deep features efficiently improved the F-Measure.
Zhong Huang and Fuji Ren : Facial expression recognition method based on multi-regional D-S evidences theory fusion, IEEJ Transactions on Electrical and Electronic Engineering (TEEE), 2016.
(Summary)
To achieve decision-level fusion of multi-regional features and highlight the credibility of different regional evidences, a facial expression recognition method based on multi-regional evidence fusion is proposed. A block histogram of gradient Gabor features in three regions, namely eyebrows, eyes, and mouth, is extracted from a facial image and regarded as evidence in expression classification. Then, category membership and regional contribution are solved with the region-weighted semisupervised fuzzy c-means clustering algorithm to construct initial basic probability assignment (BPA) and emphasize the importance of different evidences, respectively. The initial BPA of evidence is further reassigned by combining region contribution and evidence supportability to reduce evidential conflict. Finally, the final decision-level fusion of multi-regional evidences is obtained based on the Dempster Shafer (D S) combination rule. The experimental results for the Cohn Kanade expression database show that the BPA construction method based on category-membership degree and the reassignment strategy based on region contribution and evidence supportability improves the recognition rate and maintains good robustness for all types of expressions. Compared with existing decision-level fusion strategies and classification methods, the proposed recognition framework based on D S evidences theory has the advantages in recognition performance and reliability, particularly in increasing the recognition rate for expressions that are difficult to distinguish, such as fear, sadness, and disgust.
Xiaohua Wang, Xinyue Liu, Lingyun Li, Fuming Zhang, MIn Hu and Fuji Ren : GlycCompSoft: Software for Automated Comparison of Low Molecular Weight Heparins Using Top-Down LC/MS Data, PLoS ONE, Vol.11, No.12, 2016.
(Summary)
Low molecular weight heparins are complex polycomponent drugs that have recently become amenable to top-down analysis using liquid chromatography-mass spectrometry. Even using open source deconvolution software, DeconTools, and automatic structural assignment software, GlycReSoft, the comparison of two or more low molecular weight heparins is extremely time-consuming, taking about a week for an expert analyst and provides no guarantee of accuracy. Efficient data processing tools are required to improve analysis. This study uses the programming language of Microsoft Excel™ Visual Basic for Applications to extend its standard functionality for macro functions and specific mathematical modules for mass spectrometric data processing. The program developed enables the comparison of top-down analytical glycomics data on two or more low molecular weight heparins. The current study describes a new program, GlycCompSoft, which has a low error rate with good time efficiency in the automatic processing of large data sets. The experimental results based on three lots of Lovenox®, Clexane® and three generic enoxaparin samples show that the run time of GlycCompSoft decreases from 11 to 2 seconds when the data processed decreases from 18000 to 1500 rows.
Chao Li, Fuji Ren and XIN KANG : Selectional Preferences Based on Distributional Semantic Model, WSEAS Transactions on Computers, Vol.15, 258-264, 2016.
(Summary)
In this paper, we propose a approach based on distributional semantic model to the selectional preference in the verb & dobj (direct object) relationship. The distributional representations of words are employed as the semantic feature by using theWord2Vec algorithm. The machine learning method is used to build the discrimination model. Experimental results show that the proposed approach is effective to discriminate the compatibility of the object words and the performance could be improved by increasing the number of training data. By comparing the previous method, the proposed method obtain the promising results with obvious improvement. Moreover, the results demonstrate that the semantics is an universal, effective and stable feature in this task, which is consistent with our awareness of using words.
(Tokushima University Institutional Repository: 118263)
136.
Xiaohua Wang, Dengyong Hou, Min Hu and Fuji Ren : Dual-modality Emotion Recognition Model Based on Temporal-spatial LBP Moment and Dempster-Shafer Evidence Fusion, Opto-Electronic Engineering, Vol.43, No.12, 154-161, 2016.
Fuji Ren and Xiao Sun : Current Situation and Development of Intelligence Robots, ZTE, Vol.14, No.S1, 25-34, 2016.
(Summary)
Industrial intelligent robots are treated as a measure of na tionalscientificlevelandtechnologyinnovation,andalsothe important symbol of high level manufacturing, while service intelligent robots can directly affect people's daily lives. The development of artificial robots in different areas is at tracting much attention around the world. This article re views the current situation and development of Chinese and internationalintelligentrobotmarketsincludingindustrialro botsandservicerobots.Theintelligentrobottechnologyand theclassificationofrobotsarealsodiscussed.Finally,appli cations of intelligent robots in various fields are concluded andthedevelopmenttrendsandoutlookofintelligentrobots areexplored.
Fuji Ren and Zhong Huang : Automatic Facial Expression Learning Method Based on Humanoid Robot XIN-REN, IEEE Transactions on Human-Machine Systems, Vol.46, No.6, 810-821, 2016.
(Summary)
The ability of a humanoid robot to display human-like facial expressions is crucial to the natural human computer interaction. To fulfill this requirement for an imitative humanoid robot, XIN-REN, an automatic facial expression learning method is proposed. In this method, first, a forward kinematics model, which is designed to reflect nonlinear mapping relationships between servo displacement vectors and corresponding expression shape vectors, is converted into a linear relationships between the mechanical energy of servo displacements and the potential energy of feature points, based on the energy conservation principle. Second, an improved inverse kinematics model is established under the constraints of instantaneous similarity and movement smoothness. Finally, online expression learning is employed to determine the optimal servo displacements for transferring the facial expressions of a human performer to the robot. To illustrate the performance of the proposed method, we conduct evaluation experiments on the forward kinematics model and the inverse kinematics model, based on the data collected from the robot's random states as well as fixed procedures by animators. Further, we evaluate the facial imitation ability with different values of the weighting factor, according to three sequential indicators (space-similarity, time-similarity, and movement smoothness). Experimental results indicate that the deviations in mean shape and position do not exceed 6 pixels and 3 pixels, respectively, and the average servo displacement deviation does not exceed 0.8%. Compared with other related studies, the proposed method maintains better space time similarity with the performer, besides ensuring smoother trajectory for multiframe sequential imitation.
(Keyword)
movement smoothness / Expression mapping / forward kinematics model / humanoid robot / inverse kinematics model / movement similarity
Xiao Sun, Fei Gao and Fuji Ren : Mining the impact of social news on the emotions of users based on Deep Model, Journal of Chinese Information Processing, 29-32, 2016.
(Summary)
This work investigates the deep features in social news which can influence the emotions of people.Three kinds of feature compression methods are used to extract shallow features from the granularities of unigram word,bigram word and theme.The work used Support Vector Machine to select the optimal shallow features of three granularities,and the optimal F1_macro are 60.5% 62.1% 63.3% separately.The work introduced Deep Belief Network (DBN) model to train and abstract the optimal shallow features,so we can get the deep features.The optimal F1_macro of DBN3 are 61.4% 63.5% 66.1% respectively.The experimental results show that the deep features abstracted by Deep Belief Network have more semantic information and better performance than shallow features in determining the influence on people's emotions by social news.
(Keyword)
Deep Belief Nets / Restricted Boltzmann Machine / Impacts on Emotion / Social News
Xiao Sun, Zhongyuan Sun and Fuji Ren : Biomedical Named Entity Recognition Based on Deep Conditional Random Fields, Moshi Shibie yu Rengong Zhineng/Pattern Recognition and Artificial Intelligence, Vol.29, No.11, 997-1008, 2016.
Xiao Sun, Jiajin He and Fuji Ren : Ironically linguistic judgement based on hybrid neural network model via multi-feature fusion, Journal of Chinese Information Processing, Vol.30, No.6, 37-45, 2016.
142.
Xiaohua Wang, Ruijing Li, Min Hu and Fuji Ren : Occluded facial expression recognition based on the fusion of local features, Journal of Image and Graphics, Vol.21, No.11, 1473-1482, 2016.
(Summary)
To reduce the effect of partial occlusion in facial expression recognition, this paper proposes a new method of facial expression recognition based on local feature fusion. Method First, the normalized images are processed by the Gaussian filter to reduce the effect of noise. According to their different contributions in facial expression recognition, all the images are then divided into two main parts: near the eye and near the mouth. To analyze considerable structure detail, these two parts are further divided into several non-overlapping blocks. The following two patterns are used to extract the features of each sub block: the difference center-symmetric local binary pattern, which is the change of center-symmetric local binary pattern; and the gradient center-symmetric local directional pattern, which is the change of difference local directional pattern. The features are marked as two binary sequences, which are then cascaded to obtain the characteristic histogram of the sub block. The final histogram of the image is obtained by cascading the histogram of each sub block. Finally, the nearest neighbor method is used for classification. Chi-square distance is used to calculate the distance among the characteristic histograms of the testing and training images. Considering the difference of the amount of information contained in each sub block and to reduce the effect of occlusion further, information entropy is used to weigh chi-square distance adaptively. Result Three cross experiments are conducted on JAFFE and CK databases. The average recognition accuracies in random occlusion, mouth occlusion, and eye occlusion cases are 92.86%, 94.76%, and 86.19% on JAFFE database, and are 99%, 98.67%, and 99% on CK database. Conclusion In the aspect of feature extraction, our method describes the image from two aspects: one is the difference of the pixel values in the gradient direction, and the other is the difference of the edge response values between gradient directions. Accordingly, the image can be fully described. In the aspect of occlusion, image segmentation and information entropy are used to weigh chi-square distance adaptively. Thus, our method can effectively reduce the effect of occlusion. Under the same experimental conditions, experimental results show the effectiveness and superiority of the proposed method to other classical local feature extraction and occlusion handling methods.
Yu Gu, Fuji Ren and Jie Li : PAWS: Passive Human Activity Recognition Based on WiFi Ambient Signals, IEEE Internet of Things Journal, Vol.3, No.5, 796-805, 2016.
(Summary)
Indoor human activity recognition remains a hot topic and receives tremendous research efforts during the last few decades. However, previous solutions either rely on special hardware, or demand the cooperation of subjects. Therefore, the scalability issue remains a great challenge. To this end, we present an online activity recognition system, which explores WiFi ambient signals for RSSI (Received Signal Strength Indicator) fingerprint of different activities. It can be integrated into any existing WLAN networks without additional hardware support. Also it does not need the subjects to be cooperative during the recognition process. More specifically, we first conduct an empirical study to gain in-depth understanding of WiFi characteristics, e.g., the impact of activities on the WiFi RSSI. Then, we present an online activity recognition architecture that is flexible and can adapt to different settings/conditions/scenarios. Lastly, a prototype system is built and evaluated via extensive real-world experiments. A novel fusion algorithm is specifically designed based on the classification tree to better classify activities with similar signatures. Experimental results show that the fusion algorithm outperforms three other well-known classifiers (i.e., NaiveBayes, Bagging and k-NN) in terms of accuracy and complexity. Important sights and hands-on experiences have been obtained to guide the system implementation and outline future research directions.
Wei Wang, Xia Zhu, Fang Fang, Zhenlu Qin, Erhui Guo and Fuji Ren : Optimal design of 3D chip scan chains based on cores-hierarchical-placement, Journal of Electronic Measurement and Instrument, Vol.30, No.10, 1482-1489, 2016.
Jun Liu, Tangya Wang and Fuji Ren : Bad-Die Recycling Technique for Yield Enhancement of Three-Dimensional Memories, Microelectronics & Computer, Vol.33, No.10, 7-12, 2016.
Xiao Sun, Chengcheng LI and Fuji Ren : Sentiment Analysis for Chinese Microblog based on Deep Neural Networks with Convolutional Extension Features, Neurocomputing, Vol.210, 227-236, 2016.
(Summary)
Related research for sentiment analysis on Chinese microblog is aiming at the analysis procedure of posts. The length of short microblog text limits feature extraction of microblog. Tweeting is the process of communication with friends, so that microblog comments are important reference information for related post. A contents extension framework is proposed in this paper combining posts and related comments into a microblog conversation for features extraction. A novel convolutional auto encoder is adopted which can extract contextual information from microblog conversation as features for the post. A customized DNN(Deep Neural Network) model, which is stacked with several layers of RBM (Restricted Boltzmann Machine), is implemented to initialize the structure of neural network. The RBM layers can take probability distribution samples of input data to learn hidden structures for better high level features representation. A ClassRBM (Classification RBM) layer, which is stacked on top of RBM layers, is adopted to achieve the final sentiment classification label for the post. Experimental results show that, with proper structure and parameters, the performance of proposed DNN on sentiment classification is better than state of the art surface learning models such as SVM or NB, which proves that the proposed DNN model is suitable for short-length document classification with proposed feature dimensionality extension method.
Yu Gu, Lianghu Quan and Fuji Ren : AAH: accurate activity recognition of human beings using WiFi signals, Concurrency and Computation : Practice & Experience, Vol.28, No.14, 3910-3926, 2016.
(Summary)
The flourishing social networks nowadays have greatly enriched our ways of communications and thus brought people in the world much closer than ever. However, critical contexts of the traditional face-to-face communications, for example, body gestures, could be missing during the online communication, hampering the user experiences. To fill in the blank, this paper presents a passive and devices-free activity recognition system, by harvesting fingerprints of different activities from ubiquitous WiFi signals. It can be integrated into any existing WLAN networks without additional hardware supports. Also, it does not need the subjects to be cooperative during the recognition process. A prototype system is built and evaluated via extensive real-world experiments. By comparing with three state-of-the art solutions, that is, K-nearest neighbor, naive Bayes, and bagging, we show the superiority of the proposed method in terms of accuracy and complexity.
(Keyword)
accurate activity recognition / WiFi signals / fusion algorithm
Fuji Ren, Yanqiu Li, Min Hu and Liangfeng Xu : Face Recognition Method Based on Multi Features Description and Local Fusion Classification Decision, Opto-Electronic Engineering, Vol.43, No.9, 82-89, 2016.
Yiming Tang and Fuji Ren : Variable Differently Implicational Inference for R- and S-Implications, International Journal of Information Technology & Decision Making, Vol.15, No.5, 1235-1264, 2016.
(Summary)
As a generalization of the compositional rule of inference (CRI) algorithm and the fully implicational algorithm, the differently implicational algorithm of fuzzy inference not only inherit the advantages of the fully implicational algorithm, but also has stronger practicability. Then, the variable differently implicational algorithm was proposed to make the current differently implicational algorithms compose a united whole. In this paper, the variable differently implicational algorithm is further researched focusing on the fuzzy modus tollens (FMT) problem. The differently implicational principle for FMT is improved. Moreover, the unified solutions of the variable differently implicational algorithm for FMT are accomplished for R- and S-implications. Following that, as an important index of fuzzy inference, the continuity of this algorithm is analyzed for main R- and S-implications, in which excellent performance is obtained. Finally, its optimal solutions as well as inference examples are provided for several specific R- and S-implications.
(Keyword)
Fuzzy inference / fuzzy modus tollens / fuzzy implication / compositional rule of inference
Fei Gao, Xiao Sun, Kunxia Wang and Fuji Ren : Chinese micro-blog sentiment analysis based on semantic features and PAD model, 15th IEEE/ACIS International Conference on Computer and Information Science, 2016.
(Summary)
With the increasing impact of social networks, microblog becomes important carrier of information and social interaction for human beings, which contains emotional states that have important research significance. We try to analysis the microblog text with the methods of emotional vocabulary, combining domain knowledge of psychology and affective computing, continuous dimension of emotion psychology PAD model which is adopted as basis of sentiment analysis. Emotional state inherent in the text is analyzed to obtain a more accurate result and achieve purposes of emotional analysis. At the same time, to achieve emotional microblog text computability from the aspect of personal characteristics. Experimental results show that the method can improve the microblog text sentiment analysis accuracy and precision. The method is able to get a good application in the different themes and different emotional features.
Xiao Sun, Xiaoqi Peng and Fuji Ren : Detect the Emotions of Whe Public Based on Cascade Neural Network Model, 15th IEEE/ACIS International Conference on Computer and Information Science, 2016.
Xiao Sun, Chongyuan Sun, Fuji Ren, Fang Tian and Kunxia Wang : Emotional Element Detection and Tendency Judgment Based on Mixed Model with Deep Features, 15th IEEE/ACIS International Conference on Computer and Information Science, 2016.
Fuji Ren, Yu Wang and Changqin Quan : A novel factored POMDP model for affective dialogue management, Journal of Intelligent and Fuzzy Systems, Vol.31, No.1, 127-136, 2016.
(Summary)
Partially observable Markov decision process (POMDP) model has been demonstrated many times to be suited for robust spoken dialogue management. Recently, some factored representations of POMDP model are designed for specific dialogue tasks. This paper proposes a novel factored POMDP model to describe a new application of affective dialogue management. Different from existing models, the user's state space and the system's observation space are both divided into two distinct components: goal and emotion. Moreover, the system's action space is for the first time factored into two parts, i.e., goal response and emotion response, and the reward function is accordingly updated by weighted sum of the two-part rewards. An example of intelligent music player is given to explain how to apply the new model to build an affective dialogue system. Four experiments are designed to reveal the influence of key parameters on the system performance. The simulation results demonstrate the rationality and feasibility of the proposed model.
(Keyword)
Dialogue management / POMDP model / affective computing / spoken dialogue system
Xiao Sun, Ting Pan and Fuji Ren : Facial Expression Recognition Using ROI-KNN Deep Convolutional Neural Networks, ACTA AUTOMATICA SINICA, Vol.42, No.6, 883-891, 2016.
(Summary)
Deep neural networks have been proved to be able to mine distributed representation of data including image, speech and text. By building two models of deep convolutional neural networks and deep sparse recti¯er neural networks on facial expression dataset, we make contrastive evaluations in facial expression recognition system with deep neural networks. Additionally, combining region of interest (ROI) and K-nearest neighbors (KNN), we propose a fast and simple improved method called OI-KNN" for facial expression classi¯cation, which relieves the poor generalization of deep neural networks due to lacking of data and decreases the testing error rate apparently and generally. The proposed method also improves the robustness of deep learning in facial expression classi¯cation.
Fuji Ren and Chao Li : Hybrid Chinese Text Classification Approach Using General Knowledge from Baidu Baike, IEEJ Transactions on Electrical and Electronic Engineering (TEEE), Vol.11, No.5, 488-498, 2016.
(Summary)
Most of the previous studies focused on enriching text representation to address text classi cation (TC) task. However, conventional classi cation approaches with VSM (vector space model) on Chinese text study intensively only the words and their relationship in some speci c corpus/dataset but ignore the basic concept of categories and the general knowledge behind the words learned and used to recognize entities by people. This paper focuses on enriching text representation and proposes a novel approach, which complements information from the online Chinese encyclopedia Baidu Baike for Chinese TC. The similarities between every text and each concept of categories and the most related words from Baidu Baike are added to the feature space. The performance of the proposed approach is measured on the Fudan University TC corpus, which is an imbalanced Chinese dataset. In the experiments, the proposed Baidu Baike-based concept similarity approach obtains promising results when compared with a previous research and the conventional method, with macro-precision of 90.31%, recall of 75.45%, and F1 score 80.32%, which are about 0.02%, 0.15%, 0.12%, respectively, higher than the conventional method, which obviously improves the recall for some small categories while keeping precision at high level and improving the macro F1 score. Moreover, the proposed approach has good expandability, so that many other knowledge bases could be integrated and many other concepts could be referred to improve the effectiveness.
(Keyword)
Baidu Baike / Chinese / general knowledge / support vector machine / text classi cation / text representation
Xiao Sun, Jiaqi Yi and Fuji Ren : Detecting Influenza States based on Hybrid Model with Personal Emotional Factors from Social Networks, Neurocomputing, Vol.210, 257-268, 2016.
(Summary)
In this paper, we exhibit how social media data can be used to detect and analyze real-word phenomena with several data mining techniques. We investigate the real-time flu detection problem and propose a flu state detection model with personal emotional factors and semantic information (Em-Flu model). First, we extract flu-related microblog posts automatically in real-time using a hybrid model composed by Support Vector Machine with features extracted from Restricted Boltzmann Machine. In order to overcome the limitation of 140 words for posts, expect for sentiment related features, association semantic rules are also adopted as additional features, such as bag of words, negative words, degree adverbs and sentiment words dictionary. For flu state detection at specific location, we propose an unsupervised model based on personal emotional factors to figure out what state of flu in specific place. For comparison, a supervised model is also built by adopting Conditional Random Fields to decide whether a poster has ``really'' catch flu and what influenza stage the poster is in. Some statistic methods and prior rules are adopted in supervised model to get the flu state of specific locations by counting the number of microblog posts in different flu states. By considering personal emotional factors, spatial features and temporal patterns of influenza, the performance of unsupervised and supervised models are both improved. The system could tell when and where influenza epidemic is more likely to occur at certain time in specific locations. In different experiments results, the hybrid models show robustness and effectiveness than state-of-the-art unsupervised and supervised model only considering the number of posts.
(Keyword)
Influenza detection / Personal emotional factors / Transition time detection / Social web mining / Hybrid model
Sun Yan, Fuji Ren, XIN KANG and Changqin Quan : Developing a Japanese Adverb-Emotion Corpus to Investigate the Effect of Adverbs in Japanese Sentence Emotion Classification, International Journal of Advanced Intelligence (IJAI), Vol.8, No.1, 99-116, 2016.
Fuji Ren, Yanqiu Li, Liangfeng Xu, Min Hu and Xiaohua Wang : Face recognition method based on local mean pattern description and double weighted decision fusion for classification, Journal of Image and Graphics, Vol.21, No.5, 565-573, 2016.
(Keyword)
face recognition / local mean pattern / double weighted decision fusion / cloud model
Kazuyuki Matsumoto, Minoru Yoshida, Seiji Tsuchiya, Kenji Kita and Fuji Ren : Slang Analysis Based on Variant Information Extraction Focusing on the Time Series Topics, International Journal of Advanced Intelligence (IJAI), Vol.8, No.1, 84-98, 2016.
(Summary)
Recently, with increase of the number of users of Social Networking Sites, online communications have been more and more actively made, raising the possibility to use the big data on SNS for analyzing the diversity of language. Japanese language uses varieties of character types and such character types are combined and used for creating words and phrases. Therefore, it is difficult to morphologically analyze such words and phrases even though morphological analysis is a basic processing in natural language processing. Such words and phrases that are not registered in morphological analysis dictionaries are usually not defined strictly and semantic interpret for them seems to vary depending on deindividual. In this study, we chronologically analyze the topics related to slang on Twitter. In this paper, as a validation experiment, we conducted a topic analysis experiment chronogically by using the sequential tweet data, and discussed the difference of topic change according to the slang types.
Changqin Quan and Fuji Ren : Textual Emotion Recognition for Enhancing Enterprise Computing, Enterprise Information Systems, Vol.10, No.4, 422-443, 2016.
(Summary)
The growing interest in affective computing (AC) brings a lot of valuable research topics that can meet different application demands in enterprise systems. The present study explores a sub area of AC techniques textual emotion recognition for enhancing enterprise computing. Multi-label emotion recognition in text is able to provide a more comprehensive understanding of emotions than single label emotion recognition. A representation of emotion state in text is proposed to encompass the multidimensional emotions in text. It ensures the description in a formal way of the configurations of basic emotions as well as of the relations between them. Our method allows recognition of the emotions for the words bear indirect emotions, emotion ambiguity and multiple emotions. We further investigate the effect of word order for emotional expression by comparing the performances of bag-of-words model and sequence model for multi-label sentence emotion recognition. The experiments show that the classification results under sequence model are better than under bag-of-words model. And homogeneous Markov model showed promising results of multi-label sentence emotion recognition. This emotion recognition system is able to provide a convenient way to acquire valuable emotion information and to improve enterprise competitive ability in many aspects.
Lei Wang, Fuji Ren and Duoqian Miao : Multi-Label Emotion Recognition of Weblog Sentence Based on Bayesian Networks, IEEJ Transactions on Electrical and Electronic Engineering (TEEE), Vol.11, No.2, 178-184, 2016.
(Summary)
An increasing number of common users, in the Internet age, tend to express their emotions on the Web about everything they like or dislike. As a consequence, the number of all kinds of reviews, such as weblogs, production reviews, and news reviews, grows rapidly. This makes it difficult for people to understand the opinions of the reviews and obtain useful emotion information from such a huge number of reviews. Many scientists and researchers have attached more attention to emotion analysis of online information in the natural language processing field. Different from previous works, which just focused on the single-label emotion analysis, this paper takes into account rich and delicate emotions and gives special regard to multi-label emotion recognition for weblog sentences based on the Chinese emotion corpus (Ren-CECps). Using the theory of Bayesian networks and probabilistic graphical model, the latent emotion variable and topic variable are employed to find out the complex emotions of weblog sentences. Our experimental results on the multi-label emotion topic model demonstrate the effectiveness of the model in recognizing the polarity of sentence emotions.
Xiaohua Wang, Wei Huang, Chao Jin, Min Hu and Fuji Ren : Facial Expression Recognition Based on the Optimal Matching of Multi-feature and Multi-classifier, Opto-Electronic Engineering, Vol.43, No.3, 73-79, 2016.
(Keyword)
facial expression recognition / Principal Component Analysis / optimal integration of multiple classifiers / adaptive decision
Bin Zhang, Changqin Quan and Fuji Ren : Overview of Speech Synthesis in Development and Methods, Journal of Chinese Computer Systems, Vol.37, No.1, 186-192, 2016.
(Summary)
In human computer interaction the most natural and the best way to exchange is the communication by human voice.Which is mainly related to speech synthesis that is the technology of converting text to speech.This paper provides a concise but deep introduction to the development of speech synthesis and propose the problems and solutions in the development of speech synthesis.The researchers who are just entering the field of speech synthesis can stand on the shoulders of giants and have a clear and deep understanding of voice synthesis and start working with a correct judgment.In this paper first the effective methods which are already exist and mainstream in speech synthesis are overall introduced and the main idea of these methods as well as their advantages and disadvantages are described.On this basis we inspire new ideas.After that this paper illustrates the efforts respectively at home and abroad in recent years that researchers have done in the field of speech synthesis.Then objective evaluate and analyze gains and losses in synthesis technology improvements and draw the trend of synthesis technology in recent years.Finally get the prospect in speech synthesis pointing the bottleneck in development and trying to give direction to solve them.
(Keyword)
speech synthesis / HTS / text analysis / HCI
165.
Changqin Quan and Fuji Ren : Weighted high-order hidden Markov models for compound emotions recognition in text, Information Sciences, Vol.329, 581-596, 2016.
(Summary)
Emotion recognition in text has attracted a great deal of attention recently due to many practical applications and challenging research problems. In this paper, we explore an efficient identification of compound emotions in sentences using hidden Markov models (HMMs). In this problem, emotion has temporal structure and can be encoded as a sequence of spectral vectors spanning an article range. The major contributions of the research include the (i) proposal of weighted high-order HMMs to determine the most likely sequence of sentence emotions in an article. The weighted high-order HMMs take into account the impact degree of context emotions with different lengths of history; (ii) introduction of a representation of compound emotions by a sequence of binary digits, namely emotion code; (iii) development of an architecture that uses the emotions of simple sentences as part of known states in the weighted high-order hidden Markov emotion models for further recognizing more unknown sentence emotions. The experimental results show that the proposed weighted high-order HMMs is quite powerful in identifying sentence emotions compared with several state-of-the-art machine learning algorithms and the standard n-order hidden Markov emotion models. And the use of emotion of simple sentences as part of known states is able to improve the performance of the weighted n-order hidden Markov emotion models significantly.
Fuji Ren and Haitao Yu : Role-explicit query extraction and utilization for quantifying user intents, Information Sciences, Vol.329, No.1, 568-580, 2016.
(Summary)
In this study, we focus on how to extract role-explicit queries from a Chinese query log. A framework that performs role-explicit query extraction concurrently with intent role annotation is proposed. Instead of independently processing each query, the human wisdom hidden in mul-sessions is deployed to quantify the certainty of a word being the kernel-object. A sliding-window algorithm is proposed to extract role-explicit queries per mul-session. To filter the unreliable results produced by this algorithm, a pattern-based algorithm that performs a global purification is designed. The entire framework enables us to obtain a repository of sufficient role-explicit queries from a query log without human intervention. We also investigate the usefulness of aggregative role-explicit query extraction. Based a repository of role-explicit queries, we derive the richness value to quantify the richness of kernel-object oriented intents. An experimental application of the proposed framework for role-explicit query extraction to the query log SogouQ shows that: it achieves satisfactory performance. Furthermore, the richness value provides a way to capture underspecified and/or ambiguous queries, which allows selective operations to be performed depending on the nature of the queries.
(Keyword)
Kernel-object / Modifier / Richness value / Query extraction
XIN KANG, Fuji Ren and Yunong Wu : Semi-Supervised Learning of Author-Specific Emotions in Micro-Blogs, IEEJ Transactions on Electrical and Electronic Engineering (TEEE), Vol.11, No.6, 768-775, 2016.
(Summary)
Learning emotions from texts has been an active research topic in affective computing. However, the lack of reliable connection between emotions and language features has caused severely biased emotion predictions. Moreover, the author-specific patterns in emotion expression could potentially affect emotion predictions, which has never been studied. In this paper, we propose a semisupervised learning algorithm to learn emotional features from large-scaled micro-blog documents with a Bayesian network, and introduce an emotion transition factor to generate the author-specific emotion predictions. We infer the author-specific emotions in micro-blog streams through belief propagation, and learn the emotional features through an expectation maximization estimation procedure. We report results of single-label and multilabel emotion predictions on a micro-blog stream corpus, and analyze the improvements achieved by the semisupervised feature learning strategy and the incorporation of emotion transition patterns. Finally, we perform personality analysis based on the authors' emotion distribution and examine emotion distributions in the learned features.
Fuji Ren and Zhong Huang : Facial expression recognition based on AAM-SIFT and adaptive regional weighting, IEEJ Transactions on Electrical and Electronic Engineering (TEEE), Vol.10, No.6, 713-722, 2015.
(Summary)
The active appearance model (AAM), one of the most effective facial feature localization methods, is widely used in frontal facial expression recognition. However, non-frontal facial expression recognition is important in many scenarios. Thus, we propose a new method for facial expression recognition based on AAM-SIFT and adaptive regional weighting. First, multi-pose AAM templates are used for pose estimation and feature point location of the facial expression image. For effective and efficient description of these feature points, a hybrid representation, which integrates gradient direction histograms based on the descriptors of scale-invariant feature transform (SIFT) and AAM, is utilized to form AAM-SIFT features. Meanwhile, according to different expression regions, AAM-SIFT features are divided into different groups and the obtained adaptive weights by means of a regional weighted method based on the fuzzy C-means (FCM) clustering algorithm. Finally, the membership degree computed by FCM, which represents the possibility for each class, is regarded as the input feature vector for support vector machine (SVM) classifier. Extensive experiments on BU-3DFE database with six facial expressions and seven poses demonstrate the effectiveness of different types of weighting strategies and the influence of different features. Comparison with other state-of-art methods illustrates that the proposed method not only improves the recognition rates of the frontal face but also has better robustness for non-frontal facial expressions.
(Keyword)
appearance model / gradient direction histogram / scale-invariant feature transform / fuzzy c-means clustering / non-frontal facial image
Tian Chen, Xin Yi, Liuyang Zheng, Wei Wang, Huaguo Liang, Fuji Ren and Jun Liu : Low Power Deterministic Test Scheme based on Viterbi, Journal of Compter-Aided Design & Computer Graphics, Vol.28, No.5, 821-829, 2015.
(Keyword)
low power test / test data compression / block compatible encoding / Viterbi algorithm
Yan Sun, Changqin Quan, XIN KANG, Zuopeng Zhang and Fuji Ren : Customer emotion detection by emotion expression analysis on adverbs, Information Technology and Management, Vol.16, No.4, 303-311, 2015.
(Summary)
The growing interests of affective computing(AC) and its implementation in e-business demandresearchers to explore the applications of AC techniques indetecting customer emotions. Recent studies have shownthat language is powerful in conveying emotions, and agood understanding of major language features bears greatimplications for customer relationship management.However, the effect of adverbs in text emotion predictionhas only been briefly mentioned in a few related works, butnever thoroughly studied. This paper addresses the gap byinvestigating how to detect customer emotions throughanalyzing the adverbs in emotion expressions. In particular,we develop a Japanese adverb emotion corpus, analyzeemotion usage of adverbs, and further derive an adverbemotionlexicon and its rule base. Utilizing these resources,we design and perform experiments to classify emotions in sentences so as to evaluate the effectiveness of differentadverb features including adverbs, adverb emotions, andadverb emotion rules. Our experiments and analysis showthe great impact of adverbs on emotion expressions, whichcan be applied to assist e-businesses in improving customerrelated processes.
Lianghu Quan, Yu Gu, Mengni Chen and Fuji Ren : Research on pasive human activity recognition using WiFi ambient signals, Journal of University of Scinece and Technology of China, Vol.45, No.4, 308-313, 2015.
(Summary)
Although traditional k-nearest neighbor(K-NN) and bagging can recognize effectively less human activities using WiFi ambient signal, recognizing seven activities, namely, empty, walking, sitting, standing, sleeping, fallen and running, is not ideal. To further improve the performance, a new algorithm named fusion algorithm has been designed. It significantly outperform K-NN and bagging in terms of the recognition ratio in both single-subject and multi-subject experiments.
Yanqiu Li, Fuji Ren, Liangfeng Xu, Min HU and Zixi Yu : Face Recognition Based on ULBP and BP Neural Network, International Journal of Advanced Intelligence (IJAI), Vol.7, No.1, 103-114, 2015.
(Summary)
Face recognition is a research hotspot in computer vision field. Compared with the traditional feature extraction algorithms, Local Binary Pattern has many advantages, such as high accuracy, fine, light illumination invariance, etc., so it is widely used to extract facial texture feature. Firstly, we put the image evenly into blocks and constructed sub-image sets. This would be more efficient to extract feature data. Through combining with BP neural network, we make use of decision fusion and weighted fusion two ways to obtain the classification results of unknown samples. Experimental results on ORL face database show that this method is effective.
(Keyword)
LBP / BP neural network / fusion / face recognition
173.
Min Hu, Xiaoyin Huang, Fuji Ren and Xiaohua Wang : Facial Expression Recognition Based on MBP and Sparse Representation, International Journal of Advanced Intelligence (IJAI), Vol.7, No.1, 95-102, 2015.
(Summary)
Recently, sparse representation attracts widespread attention in the field of pattern recognition and artificial intelligence. Gabor features based sparse representation classifier had been introduced in face recognition research field and achieved good recognition effect. But its enormous time cost and space cost and computational complexity in the real system cannot be ignored. So this paper proposed a novel facial expression recognition method based on monogenic binary pattern and sparse representation. The proposed method extracts features from the facial expression image by using monogenic binary pattern and utilizes information entropy to weight the feature block, then constructs sparse dictionary and makes use of residual value for classification. Our proposed method was tested on the popular FERET and Cohn-Kanade facial database and obtained the average rates 94.92% and 97.46% respectively. The experimental results show that the sparse dictionary which is constructed by the proposed features has stronger ability for sparse representation and has better performance in time and space efficiency.
Jiaqi Ye, Xiao Sun, Fuji Ren and Fang Tian : Social Network Influenza Epidemic Detection Based on SVM and CRF, International Journal of Advanced Intelligence (IJAI), Vol.7, No.1, 66-79, 2015.
(Summary)
Influenza is an acute respiratory illness and widespread activity that occurs every year. Detection and prevention of Influenza in its earliest stage would reduce the spread range of the illness. Sina microblog is a popular micro blogging service in China, which could provide perfect reference sources for flu detection due to its real-time characteristic and large number of active users posting about their daily life continually. In this paper, we investigate the real-time flu detection problem and propose a flu detection model with emotion factors and semantic information (Em-Flu model). First, we extract flu-related microblog posts automatically in real-time using a simple SVM filter. We use association rule mining to extract strongly associated features as additional features of posts to overcome the limitation of 140 words for posts, including sentimental analysis information which can help to classify the posts without features. Then Conditional Random Field model is revised and applied to detect the transition time of flu that we can find out which place is more likely for influenza outbreak and when an influenza outbreak is more likely in a particular city or province in China. The experimental results display when and where influenza epidemic is more likely to occur and show the robustness and effectiveness of the proposed model might help health organizations in predicting a flu outbreak allowing them to take appropriate action in a timely manner.
(Keyword)
Influenza Detection / Conditional Random Field; / transition time detection / Social media mining / Public health
175.
Xiaohua Wang, Dengyong Hou, Min Hu and Fuji Ren : Dual-modality emotion recognition of facial expressions and gestures, International Journal of Advanced Intelligence (IJAI), Vol.7, No.1, 24-34, 2015.
(Summary)
We carry out a research of dual-modality emotion recognition about facial expressions and gestures and then propose a method combined facial expressions and gestures for dual-modality emotion recognition. Firstly, we obtain the expression image and gesture image by pre-processed the hand-over-face image. The pretreatment included reconstruct facial expression image through PCA algorithm. Secondly, we extract the LBP features from expression image and the Hu moment invariant features and Fourier descriptor features from gesture image. We use SVM and KNN to coarse classification recognition the expression features and gesture features respectively. Finally, we mix classification results of facial expressions and gestures together by using the linear weighted voting algorithm proposed in this paper. The experimental results show that this method has the better discrimination in the dual-modality emotional database established by ourselves.
Min Hu, Yihong Cheng, Xiaohua Wang, Fuji Ren, Liangfeng Xu and Xiaoyin Huang : Facial Expression Recognition Based on asymmrtric region local gradient coding, Journal of Image and Graphics, Vol.20, No.10, 1313-1321, 2015.
Xiao Sun, Chongyuan Sun and Fuji Ren : New Word Detection and Emotional Tendency Judgment Based on Deep Structured Model, Computer Science, Vol.42, No.9, 208-213, 2015.
(Summary)
With the development of social network,new words appear ceaselessly.The appearance of new word tends to characterize the social hot spot or represent certain public mood.The new word detection and emotional tendency judgment provide a new way for the public mood forecast.We constructed the deep conditional random fields model for the sequence labeling,introduced part of speech,character position,the ability of word formation as features,and combined it with the crowd sourcing network dictionary and the other third party dictionary.Traditional method based on emotional dictionary is difficult to judge the new word emotional tendency.We expressed word as a vector of K dimension based on neural network language model in order to find the nearest words to the new word in the vector space.According to the emotional tendency of these words and the distance between them and the new word,the new word sentiment is judged.The experiment on corpus of Peking university demonstrates the feasibility of the proposed model and method,in which the new word detection F-value is 0.991,and the emotion recognition accuracy is 70%.
Yiming Tang and Fuji Ren : Variable differently implicational algorithm of fuzzy inference, Journal of Intelligent and Fuzzy Systems, Vol.28, No.4, 1885-1897, 2015.
(Summary)
To reveal the inherent essence of current differently implicational algorithms for fuzzy inference, the variable differently implicational algorithm is put forward and investigated. The differently implicational principles are improved from variable and generalized viewpoint. Furthermore, focusing on the FMP (fuzzy modus ponens) problem, unified forms of the new algorithm are obtained for R-implications and S-implications. Following that, the optimal differently implicational solutions are achieved for several specific R-implications and S-implications. Lastly, the new algorithm makes the current differently implicational algorithms compose a united whole.
(Keyword)
Fuzzy inference / fuzzy modus ponens / fuzzy implication / compositional rule of inference / fully implicational algorithm
Ye Wu and Fuji Ren : Exploiting Opinion Distribution for Topic Recommendation in Twitter, IEEJ Transactions on Electrical and Electronic Engineering (TEEE), Vol.10, No.5, 567-575, 2015.
(Summary)
The popular social networking service Twitter provides rapidly updated information and online trends, which enriches andbenefits peoples daily life. At the same time, how to find out the really interesting and relevant topics from the massivestreams of tweets, to provide precise topic recommendation for users, becomes a challenging problem in the real world. Previouscollaborative filtering methods give solutions to traditional recommendation tasks considering users positive reviews to helprecommend items. However, to the problem what is interesting to whom in Twitter, positive opinions toward a topic do notimply that the user will be interested in it with high probability, for the user probably prefers to know those controversial topicsor hot events with a large number of negative posts. In this paper, we exploit the characteristics of topical opinion distributionfor improving the performance of recommendation. The experimental results on a real-world Twitter dataset show that theproposed opinion-distribution-aware topic recommendation (ODA-TR) approach outperforms the state-of-the-art collaborativerecommendation methods
(Keyword)
collaborative filtering / opinion distribution / topic recommendation / Twitter
Fuji Ren and Kazuyuki Matsumoto : Semi-automatic Creation of Youth Slang Corpus and Its Application to Affective Computing, IEEE Transactions on Affective Computing, Vol.7, No.2, 176-189, 2015.
(Summary)
This paper proposes a method to semi-automatically construct a corpus that includes Japanese youth slang called Wakamono Kotoba. The process of semi-automatic corpus construction is composed of the first step is automatic collection of example sentence, the second step is tag annotation to collected sentences, and the final step is manually modifying tag and noise reduction. In this process, there are two problems. The first problem is quality of the automatic collected corpora. The second is the accuracy of tag annotation. If the automatically annotated tags are unreliable, after all, it takes long time to modify them manually. As a solution of the first problem, we proposed a filtering method to remove meaningless sentences (noise sentences) automatically. In order to solve a second problem, we proposed an emotion estimation method that can be applied to the sentences that included youth slang and were difficult to be analyzed automatically. The result of the accuracy evaluation showed improvement in F1-Score compared to the machine learning method and confirms the effectiveness of the proposed method.
(Keyword)
Emotion corpus / semi-automatic corpus construction / Affective Computing / youth slang / Wakamono Kotoba
Fuji Ren, XIN KANG and Changqin Quan : Examining Accumulated Emotional Traits in Suicide Blogs with an Emotion Topic Model, IEEE Journal of Biomedical and Health Informatics, Vol.20, No.5, 1384-1396, 2015.
(Summary)
Suicide has been a major cause of death throughout the world. Recent studies have proved a reliable connection between the emotional traits and suicide. However, detection and prevention of suicide are mostly carried out in the clinical centers, which limits the effective treatments to a restricted group of people. To assist detecting suicide risks among the public, we propose a novel method by exploring the accumulated emotional information from people's daily writings (i.e. Blogs), and examining these emotional traits which are predictive of suicidal behaviors. A complex emotion topic (CET) model is employed to detect the underlying emotions and emotion-related topics in the Blog streams, based on eight basic emotion categories and five levels of emotion intensities. Since suicide is caused through an accumulative process, we propose three accumulative emotional traits, i.e., accumulation, covariance, and transition of the consecutive Blog emotions, and employ a generalized linear regression algorithm to examine the relationship between emotional traits and suicide risk. Our experiment results suggest that the emotion transition trait turns to be more discriminative of the suicide risk, and that the combination of three traits in linear regression would generate even more discriminative predictions. A classification of the suicide and non-suicide Blog articles in our additional experiment verifies this result. Finally, we conduct a case study of the most commonly mentioned emotion-related topics in the suicidal Blogs, to further understand the association between emotions and thoughts for these authors.
Min Hu, Kun Li, Xiaohua Eang and Fuji Ren : Facial expression recognition based on histogram weighted HCBP, Journal of Electronic Measurement and Instrumentation, Vol.29, No.7, 953-960, 2015.
(Summary)
In order to overcome the limitation of local binary pattern ( LBP) and its improved algorithm a facial expression method based on histogram weighted HCBP is proposed Firstly facial image is divided into some uniform sub-image and HCBP operator is used to extract texture feature Then the information entropy is used to calculate the weight of every sub-image weighted HCBP histogram of sub-image is combined with the HCBP histogram of the original image and the result histogram image is accomplished as the facial expression feature Finally the expression is classified with the nearest neighbor classifier Using the combination of Haar-like feature and CBP operator makes the description of local feature more sufficient The introduction of information entropy can distinguish the contribution of different partitions of the face The experimental results in JAFFE library and Cohn-Kanade library show that the HCBP method outperforms than existing LBP methods in both the recognition rate and the speed
Bin Zhang, Fuji Ren and Changqin Quan : Overview of Speech Synthesis in Development and Methods, Journal of Chinese Computer Systems, Vol.36, No.6, 1-8, 2015.
(Summary)
Spoken dialogue system is the core technology in the field of human-computer interaction, and it is an important way to realize the harmonious human-computer interaction, the research has great theory significance and application value. The advances of theory and technology in spoken dialogue systems have always been greatly concerned. The status and advances of dialogue management and spoken dialogue system are comprehensively summarized in this paper. First, the main research questions of spoken dialogue system are introduced comprehensively, including the research contents of the modules, key technologies, portability and robust design. Then, the various spoken dialogue management strategies are systematically analyzed from the perspective of theoretical models, advances and usability. Finally, several possible directions and problems for further consideration and discussion are also mentioned.
184.
Yu Gu and Fuji Ren : Energy-efficient Indoor Localization of Smart Hand-held Devices Using Bluetooth, IEEE Access, Vol.3, 1450-1461, 2015.
(Summary)
Indoor localization of smart hand-held devices is essential for location based services of pervasive applications. Previous research mainly focuses on exploring wireless signal fingerprints for this purpose, and several shortcomings need to be addressed first before real-world usage, e.g., demanding a large number of APs or labor-intensive site survey. In this paper, through a systematic empirical study, we first gain in-depth understandings of Bluetooth characteristics, i.e., the impact of various factors such as distance, orientation, and obstacles on the Bluetooth RSSI (Received Signal Strength Indicator). Then, by mining from historical data, a novel localization model is built to describe the relationship between RSSI and the device location. On this basis, we present an energy- efficient indoor localization scheme that leverages user motions to iteratively shrink the search space to locate the target device. MLDB has been prototyped and evaluated in several real-world scenarios. Extensive experiments show that our algorithm is efficient in terms of localization accuracy, searching time and energy consumption.
(Keyword)
Energy efficiency / indoor Localization; / Data Mining / Bluetooth / IoT
Fuji Ren, Mengni Chen and Yu Gu : WeWatch: An Application for Watching Video Across Two Mobile Devices, ZTE COMMUNICATIONS, Vol.13, No.2, 17-22, 2015.
(Summary)
In recent years, high resolution video has developed rapidly and widescreen smart devices have become popular. We present an Android application called WeWatch that enables high resolution video to be shared across two mobile devices when they are close to each other. This concept has its inspiration in machine to machine connections in the Internet of Things (loT). We ensure that the two parts of the video are the same size over both screens and are synchronous. Further, a user can play, pause, or stop the video by moving one device a certain distance from the other. We decide on appropriate distances through experimentation. We implemented WeWatch on Android operating system and then optimize Watch so battery consumption is reduced. The user ex perience provided by WeWatch was evaluated by students through a questionnaire, and the reviews indicated that WeWatch does improve the viewing experience.
(Keyword)
together watching experience / screen adaptation / internet of things / distance estimation / energy efficiency
Jun Liu, Qingqing Qian, Xi Wu, Wei Wang, Tian Chen and Fuji Ren : Optimizing pre-bond and post-bond test time for three dimension IP Cores, Computer Engineering and Applications, Vol.19, No.11, 1-7, 2015.
Fuji Ren, Yu Wang and Changqin Quan : TFSM-based dialogue management model framework for affective dialogue systems, IEEJ Transactions on Electrical and Electronic Engineering (TEEE), 1-7, 2015.
(Summary)
A new dialogue management model for affective dialogue system, which aims to provide a service of information inquiry and affective interaction, is proposed in this paper. First, we construct two finite state machines (TFSM) to model the user and the system, respectively, and simulate the dialogue process as an information exchange between the two state machines. All possible state transitions in dialogue and its probabilities of the user are summarized as a user model, which is helpful for the system to inference and predict the user's internal states. Second, we further discuss the implementation methods of information inquiry and emotional response modules. Finally, we employ the return function of partially observable Markov decision processes (POMDP) model to analyze and evaluate the TFSM-based dialogue management model. The experimental results not only show the relationships between the average returns, recognition error rates, and state transition probabilities but also confirm that our TFSM-based dialogue management model outperforms the conventional FSM model.
(Keyword)
dialogue management model / affective dialogue system / finite state machine / affective computing
Min Hu, Jiang Hr, Xiaohua Wang, Hongbo Chen, Kun Li and Fuji Ren : Precise local feature description for facial expression recognition, Journal of Image and Graphics, Vol.19, No.11, 1613-1622, 2015.
(Keyword)
acial expression recognition / precise local features / sufficient vector triangle / a variety of scales
Yu Gu, Fuji Ren, Yusheng JI and Ji Li : The Evolution of Sink Mobility Management in Wireless Sensor Networks: A Survey, IEEE Communications Surveys and Tutorials, 1-19, 2015.
(Summary)
Sink mobility has long been recognized as an efficient method of improving system performance in wireless sensor networks (WSNs), e.g. relieving traffic burden from a specific set of nodes. Though tremendous research efforts have been devoted to this topic during the last decades, yet little attention has been paid for the research summarization and guidance. This paper aims to fill in the blank and presents an up-to-date survey on the sink mobility issue. Its main contribution is to review mobility management schemes from an evolutionary point of view. The related schemes have been divided into four categories: uncontrollable mobility (UMM), pathrestricted mobility (PRM), location-restricted mobility (LRM) and unrestricted mobility (URM). Several representative solutions are described following the proposed taxonomy. To help readers comprehend the development flow within the category, the relationship among different solutions is outlined, with detailed descriptions as well as in-depth analysis. In this way, besides some potential extensions based on current research, we are able to identify several open issues that receive little attention or remain unexplored so far.
Changqin Quan and Fuji Ren : Feature-level sentiment analysis by using comparative domain corpora, Enterprise Information Systems, 1-18, 2014.
(Summary)
Feature-level sentiment analysis (SA) is able to provide more fine-grained SA oncertain opinion targets and has a wider range of applications on E-business. Thisstudy proposes an approach based on comparative domain corpora for feature-levelSA. The proposed approach makes use of word associations for domain-specificfeature extraction. First, we assign a similarity score for each candidate feature todenote its similarity extent to a domain. Then we identify domain features based ontheir similarity scores on different comparative domain corpora. After that, dependencygrammar and a general sentiment lexicon are applied to extract and expand feature orientedopinion words. Lastly, the semantic orientation of a domain-specific feature isdetermined based on the feature-oriented opinion lexicons. In evaluation, we comparethe proposed method with several state-of-the-art methods (including unsupervised andsemi-supervised) using a standard product review test collection. The experimentalresults demonstrate the effectiveness of using comparative domain corpora.
(Keyword)
affective computing / affective computing / frequencyinverse document frequency / pointwise mutual information / pointwise mutual information
Jun Liu, Xi Wu, Huaguo Liang and Fuji Ren : Optimizing the number of leaf nodes and TSVs in three dimensional scan tree, SCIENTIA SINICA Informationis, Vol.44, No.1, 1-13, 2014.
(Summary)
Scan tree architecture can effectively reduce test data volume, test time and test cost for integratedcircuits. To reduce the number of leaf nodes and TSVs(through silicon vias) in scan tree for three dimensionalintegrated circuits, this paper firstly draws the conclusion that the minimum number of leaf nodes is the numberof scan cells contained in the maximal compatible group. Then, the necessary and sufficient condition achievingthe minimum number of leaf nodes is presented. On the basis above, a heuristic algorithm is proposed, whichcan minimize the number of leaf nodes and reduce consumed TSVs as many as possible. Experimental resultsdemonstrate the effectiveness of the proposed technique.
Changqin Quan and Fuji Ren : Visualizing Emotions from Chinese Blogs by Textual Emotion Analysis and Recognition Techniques, International Journal of Information Technology & Decision Making, Vol.13, 1-20, 2014.
(Summary)
The research on blog emotion analysis and recognition has become increasingly important in recent years. In this study, based on the Chinese blog emotion corpus (Ren-CECps), we analyze and compare blog emotion visualization from different text levels: word, sentence, and paragraph. Then, a blog emotion visualization system is designed for practical applications. Machine learning methods are applied for the implementation of blog emotion recognition at different textual levels. Based on the emotion recognition engine, the blog emotion visualization interface is designed to provide a more intuitive display of emotions in blogs, which can detect emotion for bloggers, and capture emotional change rapidly. In addition, we evaluated the performance of sentence emotion recognition by comparing five classification algorithms under different schemas, which demonstrates the effectiveness of the Complement Naive Bayes model for sentence emotion recognition. The system can recognize multi-label emotions in blogs, which provides a richer and more detailed emotion expression.
Jun Wang, Fuji Ren and Lei Li : Recognizing Sentiment of Relations between Entities in Text, IEEJ Transactions on Electrical and Electronic Engineering (TEEE), Vol.9, No.6, 614-620, 2014.
Xiaohua Wang, Chao Jin, Fuji Ren and Min Hu : Research on facial expression recognition based on pyramid Weber local descriptor and the Dempster-Shafer theory of evidence, Journal of Image and Graphics, Vol.19, No.9, 1297-1305, 2014.
(Summary)
Feature extraction is the most critical step in pattern recognition, and facial expression recognition is no exception.Weber Local Descriptor (WLD) is a Method that can effectively extract texture information from images and has the advantages of being consistent with human perception of human beings and being insensitive to noise and non-monotonic illumination variations. However, WLD has some limitations in the feature representation of local details. To overcome these limitations, a facial expression recognition Method based on Pyramid WLD (PWLD) is proposed in this study.Method First, facial images are preprocessed. This step includes the detection of faces from facial expression databases and normalization. The salient regions of segment 2 that have significant contributions to facial expression recognition from images are also preprocessed. One of these salient regions is that which includes the eyes and eye brows, while another is that with the mouth. The sizes of salient regions differ, and these regions contain different information. Thus, we stratify these salient regions and divide each layer into different blocks. The PWLD features of each block in each layer are then extracted and cascaded to represent the global and local features of a salient region reasonably, with some parameter adjustments. Second, we compute for the Chi-square distance of the PWLD histograms in both the testing and training sets. We then choose the minimum distance in every category of expressionsand normalize this distance to construct the Basic Probability Assignment (BPA) as independent evidence. To create the BPA, we use curve fitting in numerical analysis by simulating several sets of data. Finally, fusion BPA is obtained by using the Dempster-Shafer rule, and the Results are further obtained by employing thedecision-making and judgment of Dempster-Shafertheory of evidence.Result By fusing the PWLD features of the two different salient regions with Dempster-Shafer theory of evidence, we can overcome the limitations of a single regional featureand acquire more reliable and accurate Results. We conduct some cross-validation experiments on the JAFFE and Cohn-Kanade facialexpression databases, and the average recognition rates reach up to 95.81% and 97.47%, respectively. In addition, we perform some experiments with other algorithms, such as LBP, LDP, and Gabor; we also conduct some comparative experiments that combine the PWLD with different classifiers, such as 1-NN and SVM.Conclusion The WLD, which is known as a robust image descriptor, can well extract the texture information of images. Moreover, the PWLD can accurately describe the local details, which have more advantages than the WLD features. The comparative Results of some typical Methodsverify the effectiveness and fault tolerance of the proposed Method. The proposed Method has certain robustness under simultaneous noise and light conditions.
(Keyword)
facial expression recognition / Weber local descriptor / pyramid Weber local descriptor / Dempster-Shafer(D-S) theory of evidence
Changqin Quan, Meng Wang and Fuji Ren : An Unsupervised Text Mining Method for Relation Extraction from Biomedical Literature, PLoS ONE, Vol.9, No.7, e102039, 2014.
(Summary)
The wealth of interaction information provided in biomedical articles motivated the implementation of text mining approaches to automatically extract biomedical relations. This paper presents an unsupervised method based on pattern clustering and sentence parsing to deal with biomedical relation extraction. Pattern clustering algorithm is based on Polynomial Kernel method, which identifies interaction words from unlabeled data; these interaction words are then used in relation extraction between entity pairs. Dependency parsing and phrase structure parsing are combined for relation extraction. Based on the semi-supervised KNN algorithm, we extend the proposed unsupervised approach to a semi-supervised approach by combining pattern clustering, dependency parsing and phrase structure parsing rules. We evaluated the approaches on two different tasks: (1) Proteinprotein interactions extraction, and (2) Genesuicide association extraction. The evaluation of task (1) on the benchmark dataset (AImed corpus) showed that our proposed unsupervised approach outperformed three supervised methods. The three supervised methods are rule based, SVM based, and Kernel based separately. The proposed semi-supervised approach is superior to the existing semi-supervised methods. The evaluation on genesuicide association extraction on a smaller dataset from Genetic Association Database and a larger dataset from publicly available PubMed showed that the proposed unsupervised and semi-supervised methods achieved much higher F-scores than co-occurrence based method.
Changqin Quan and Fuji Ren : Unsupervised Product Feature Extraction for Feature-oriented Opinion Determination, Information Sciences, Vol.272, 16-28, 2014.
(Summary)
Identifying product features from reviews is the fundamental step as well as a bottleneck in feature-level sentiment analysis. This study proposes a method of unsupervised product feature extraction for feature-oriented opinion determination. The domain-specific features are extracted by measuring the similarity distance of domain vectors. A domain vector is derived based on the association values between a feature and comparative domain corpora. A novel term similarity measure (PMITFIDF) is introduced to evaluate the association of candidate features and domain entities. The results show that our approach of feature extraction outperforms other state-of-the-art methods, and the only external resources used are comparative domain corpora. Therefore, it is generic and unsupervised. Compared with traditional pointwise mutual information (PMI), PMITFIDF showed better distinction ability. We also propose feature-oriented opinion determination based on feature-opinion pair extraction and feature-oriented opinion lexicon generation. The results demonstrate the effectiveness of our proposed method and indicate that feature-oriented opinion lexicons are superior to general opinion lexicons for feature-oriented opinion determination.
Misao Miyagawa, Tetsuya Tanioka, Yuko Yasuhara, Kazuyuki Matsumoto, Hirokazu Ito, Motoyuki Suzuki, Fuji Ren and Rozzano De Castro Locsin : Methodology for Developing a Nursing Administration Analysis System, Intelligent Information Management, Vol.6, No.3, 118-128, 2014.
(Summary)
Nursing administration requires a large volume of wide-ranging information, and nurse administrators are limited in their ability to compile and analyze information for nursing administration. The purpose of this study is to create methodology for developing a nursing administration analysis system to aid nurse administrators in performing outcome analysis. In this methodology, information required for nursing administration in the PSYCHOMS? (Psychiatric Outcome Management System, registered trademark) database is analyzed according to the individual needs of nurse administrators. It features a combination of a classification method and an extraction method for obtaining quantitative and qualitative data as information required for nursing administration, and enables nurse administrators to easily obtain analysis results that they directly need. This methodology converts the time required nurse administrators to collect and organize information into time for making considerations in order to devise strategies for improving the quality of nursing care services, and can improve the quality and efficiency of nursing administration. This may lead to an increase of the quality of nursing care services at psychiatric hospitals. This methodology is highly versatile as it can be applied in information management, not only for nursing, but for the entire psychiatric hospital.
(Keyword)
Nursing Administration / Clinical Pathways / analysis of variance / Analysis System / PSYCHOMS®
Changqin Quan and Fuji Ren : Target Based Review Classification for Fine-Grained Sentiment Analysis, International Journal of Innovative Computing, Information and Control, Vol.10, No.1, 257-268, 2014.
(Summary)
Target based sentiment classification is able to provide more fine grained sentiment analysis. In this paper, we propose a similarity based approach for this problem. Firstly, a new measure of PMI-TFIDF by combining PMI (Pointwise mutual information) and TF-IDF (term frequencyinverse document frequency) is proposed to measure the association of words for extending related features for a given target. Then Polynomial Kernel (PK) method is applied to get the similarities between a review and the related features of different targets. The sentiment orientation of a review is determined by comparing their similarities with the target based opinion words. The comparisons between PMI and PMI-TFIDF showed that the extracted features that measured by PMI-TFIDF have closer association with the targets than the extracted features measured by PMI. And the association values measured by PMI-TFIDF showed better distinction between different features. The experiments also demonstrated the effectiveness and validation of the proposed approach on target based review classification, opinion words extraction, and target based sentiment classification.
Haotao Yu and Fuji Ren : Automatic Role-explicit Query Extraction: A Divide-and-Conquer System Leveraging on Users' Reformulating Behaviors, IEEJ Transactions on Electrical and Electronic Engineering (TEEE), Vol.9, No.1, 62-70, 2014.
(Summary)
This paper presents a system that can automatically extract role-explicit queries from a query log without any human intervention. The key idea underlying our system is as follows: We perform a divide-and-conquer process through differentiating the sessions in a query log as mul-sessions and sin-sessions. According to the session type, different approaches are proposed. We translate the contextual information in mul-sessions as indirect human wisdom to facilitate role-explicit query extraction on mul-sessions. Furthermore, leveraging on the role-explicit queries extracted from mul-sessions, we learn the simplified word n-gram role model (SWNR) to facilitate role-explicit query extraction on sin-sessions. The experimental results show that our proposed system is clearly favored by the indirect human wisdom hidden in mul-sessions and achieves a satisfactory performance, namely more than 79% in terms of different metrics.
(Keyword)
role-explicit / intent role / kernel-object / modifier
Fuji Ren and Ye Wu : Predicting User-topic Opinions in Twitter with Social and Topical Context, IEEE Transactions on Affective Computing, Vol.4, No.4, 412-424, 2013.
(Summary)
With popular microblogging services like Twitter, users are able to share opinions towards hot topics online, in a more convenient way. The user generated data in Twitter is thus regarded as an opinion-rich resource, and has been exploited for various applications. Most existing work mines opinions in Twitter by performing sentiment analysis on the text of tweets. However, a user's opinion towards a topic cannot be obtained by using those text-based approaches, when the user hasn't posted any tweet about it yet. In some cases such as launching new products or proposing new policies, the related online comments are unavailable before their release, while inferring users' opinions towards them beforehand could be potentially helpful for making correct decisions. Therefore, how to predict a user's opinions towards topics without observing the corresponding tweets from him/her, becomes a novel problem presenting both challenges and opportunities. We focus on the problem of predicting user-topic opinions, and propose a framework mph{ScTcMF} for incorporating social and topical context to solve it. The experimental results on a real-world Twitter dataset show that our framework outperforms the state-of-the-art collaborative filtering methods, and demonstrate that both social context and topical context are effective in improving the prediction performance.
Fuji Ren, Bo Li and Qimei Chen : Single Parameter Logarithmic Image Processing For Edge detection, IEICE Transactions on Information and Systems, Vol.E96-D, No.11, 2437-2449, 2013.
(Summary)
Considering the non-linear properties of the human visual system, many non-linear operators and models have been developed, particularly the logarithmic image processing (LIP) model proposed by Jourlin and Pinoli, which has been proved to be physically justified in several laws of the human visual system and has been successfully applied in image processing areas. Recently, several modifications based on this logarithmic mathematical framework have been presented, such as parameterized logarithmic image processing (PLIP), pseudo-logarithmic image processing, homomorphic logarithmic image processing. In this paper, a new single parameter logarithmic model for image processing with an adaptive parameter-based Sobel edge detection algorithm is presented. On the basis of analyzing the distributive law, the subtractive law, and the isomorphic property of the PLIP model, the five parameters in PLIP are replaced by a single parameter to ensure the completeness of the model and physical constancy with the nature of an image, and then an adaptive parameter-based Sobel edge detection algorithm is proposed. By using an image noise estimation method to evaluate the noise level of image, the adaptive parameter in the single parameter LIP model is calculated based on the noise level and grayscale value of a corresponding image area, followed by the single-parameter LIP-based Sobel operation to overcome the noise-sensitive problem of classical LIP-based Sobel edge detection methods, especially in the dark area of an image, while retaining edge sensitivity. Compared with the classical LIP and PLIP model, the given single parameter LIP achieves satisfactory results in noise suppression and edge accuracy.
Yimin Tang and Fuji Ren : Universal Triple I Method for Fuzzy Reasoning and Fuzzy Controller, Iranian Journal of Fuzzy Systems, Vol.10, No.5, 1-24, 2013.
(Summary)
As a generalization of the triple I method, the universal triple Imethod is investigated from the viewpoints of both fuzzy reasoningand fuzzy controller. The universal triple I principle is putforward, which improves the previous triple I principle. Then,unified form of universal triple I method is established based onthe (0,1)-implication or R-implication. Moreover, the reversibilityproperty of universal triple I method is analyzed from expansion,reduction and other type operators, which demonstrate that itsreversibility property seems fine, especially for the case employingthe (0,1)-implication. Lastly, we analyze the response ability offuzzy controllers based on universal triple I method, then thepracticability of triple I method is improved.
Kunxia Wang, Lian Li, Jing Yang and Fuji Ren : Speech Emotion Recognition Using a Novel Feature Set, Journal of Computational Information Systems, Vol.9, No.15, 6097-6104, 2013.
(Summary)
Speech emotion recognition has attracted attentions from increased number of researchers in Psychology,Computer science,Phonetics and related disciplines.This paper discusses combining multiple features forspeech emotion recognition to yield improved performance. It proposes using combination of multiplefeatures to improve recognition performance. Using ChineseLDC emotional database, the experimentresults show that a novel feature set of MFCC, fundamental frequency, MFCC, MFCC, energy,zero-crossing ratio and amplitude can get the best performance as 95% and 84% for emotional speechrecognition of four emotional state and six emotional state.
(Keyword)
Speech Emotion Recognition / Support Vector Machine / Feature Set
Ji Li and Fuji Ren : Hybrid Approach for Word Emotion Recognition, IEEJ Transactions on Electrical and Electronic Engineering (TEEE), Vol.8, No.6, 616-626, 2013.
(Summary)
In recent years, with plenty of online resources constantly emerging, emotion recognition in text has become increasingly important in human-computer interaction. Word emotion plays a very important role in emotion analysis of sentences or documents. This paper proposes a hybrid approach to recognize word emotion in the dimension of eight emotion categories with corresponding intensities based on the Chinese emotion corpus. First, we present a new algorithm of semantic similarity computation for aiding emotion intensity computation, and design a new algorithm of emotion vector computation by making use of both morpheme characteristics and semantic relations. And then, we adopt Support Vector Machines (SVM) model for the secondary classification to the words whose emotions can not be calculated by the semantic analysis algorithm. Our approach achieves the accuracy of 54.00% and 78.75% for exact match and all five types of hit respectively on the basis of the core emotion lexicon CL4. Experimental results show that the integration of morpheme characteristics and semantic relations can improve the classification accuracy efficiently.
Degen HUANG, Shanshan WANG and Fuji Ren : Creating Chinese-English Comparable Corpora, IEICE Transactions on Information and Systems, Vol.E96-D, No.8, 1853-1861, 2013.
(Summary)
Comparable Corpora are valuable resources for many NLP applications, and extensive research has been done on information mining based on comparable corpora in recent years. While there are not enough large-scale available public comparable corpora at present, this paper presents a bi-directional CLIR-based method for creating comparable corpora from two independent news collections in different languages. The original Chinese document collections and English documents collections are crawled from XinHuaNet respectively and formatted in a consistent manner. For each document from the two collections, the best query keywords are extracted to represent the essential content of the document, and then the keywords are translated into the language of the other collection. The translated queries are run against the collection in the same language to pick up the candidate documents in the other language and candidates are aligned based on their publication dates and the similarity scores. Results show that our approach significantly outperforms previous approaches to the construction of Chinese-English comparable corpora.
(Keyword)
Comparable Corpora / cross language information retrieval / keyword extraction / document alignment
Fuji Ren, Changqin Quan and Kazuyuki Matsumoto : ENRICHING MENTAL ENGINEERING, International Journal of Innovative Computing, Information and Control, Vol.9, No.8, 3271-3286, 2013.
(Summary)
The new growing research eld of Affective Computing (AC) provides a newhorizon for quantitative analysis of human emotional states using IT techniques. In thispaper, a new academic system called nriching Mental Engineering (EME)" is proposedfor the problem of mental health from the view of engineering. EME is being establishedas an academic discipline, by being keenly aware of the poverty of the mind from whichpeople living in modern society suffer. In EME, quantitative measurement of richness ofthe mind is regarded as a central technique. This is measured from the information likesubject physiological data, textual information, behavior, and tone of voice. Meanwhile,EME also systemizes external stimuli by an emotional energy function. The emotionenergy function is proposed to calculate a person's emotional stimuli at a certain pointfrom factors like choice of words, voice, facial expressions, physiological information, andbehavior. Furthermore, an application of EME is illustrated through an analysis of thedepressive tendencies in blogs.
Yu Gu, Yusheng Ji, Ji Li, Fuji Ren and Baohua Zhao : EMS: Efficient Mobile Sink Scheduling in Wireless Sensor Networks, Ad Hoc Networks, Vol.11, No.5, 1556-1570, 2013.
(Summary)
Sink scheduling, in the form of scheduling multiple sinks among the available sink sites to relieve the level of traffic burden, is shown to be a promising scheme in wireless sensor networks (WSNs). However, the problem of maximizing the network lifetime via sink scheduling remains quite a challenge since routing issues are tightly coupled. Previous approaches on this topic either suffer from poor performance due to a lack of joint considerations, or are based on relaxed constraints. Therefore, in this paper, we aim to fill in the research blanks. First, we develop a novel notation Placement Pattern (PP) to bound time-varying routes with the placement of sinks. This bounding technique transforms the problem from time domain into pattern domain, and thus, significantly decreases the problem complexity. Then, we formulate this optimization in a pattern-based way and create an efficient Column Generation (CG) based approach to solve it. Simulations not only demonstrate the efficiency of the proposed algorithm but also substantiate the importance of sink mobility for energy-constrained WSNs.
Elmarhoumy Mahmoud, Fattah Abdel Mohamed, Motoyuki Suzuki and Fuji Ren : A New Modified Centroid Classifier Approach for Automatic Text Classification, IEEJ Transactions on Electrical and Electronic Engineering (TEEE), Vol.8, No.4, 364-370, 2013.
(Summary)
To enhance the automatic text classification task, this paper proposes a novel approach for treating the problem of inductive bias incurred by the centroid classifier assumption. This approach is a trainable classifier, which takes into account tfidf as a text feature. The main goal of the proposed approach is to take advantage of the most similar training errors in the classification model for successively updating that model based on a certain threshold. The proposed approach is practical and flexible to implement. The complete performance of the proposed approach is measured at several threshold values on the Reuters-21578 text categorization collection. Experimental results show that the proposed approach can improve the performance of the centroid classifier better than traditional approaches (traditional centroid classifier, support vector machines, decision trees, Bayes nets, and N Bayes) by 1, 1.2, 4.1, 7.5, and 11%, respectively.
(Keyword)
Text classification / text categorization / centroid classifier / vector space model / modified centroid classifier
Fuji Ren and XIN KANG : Employing Hierarchical Bayesian Networks in Simple and Complex Emotion Topic Analysis, Computer Speech & Language, Vol.27, No.4, 943-968, 2013.
(Summary)
Traditional emotion models, when tagging single emotions in documents, often ignore the fact that most documents convey complex human emotions. In this paper, we join emotion analysis with topic models to find complex emotions in documents, as well as the intensity of the emotions, and study how the document emotions vary with topics. Hierarchical Bayesian networks are employed to generate the latent topic variables and emotion variables. On average, our model on single emotion classification outperforms the traditional supervised machine learning models such as SVM and Naive Bayes. The other model on the complex emotion classification also achieves promising results. We thoroughly analyze the impact of vocabulary quality and topic quantity to emotion and intensity prediction in our experiments. The distribution of topics such as Friend and Job are found to be sensitive to the documents' emotions, which we call emotion topic variation in this paper. This reveals the deeper relationship between topics and emotions.
Elmarhoumy Mahmoud and Fuji Ren : A New Hybrid Model for Automatic Text Classification, The Online Journal on Computer Science and Information Technology, OJCSIT, Vol.3, No.2, 132-137, 2013.
(Summary)
This paper introduces three novel pointes the first one is creating the proposed tfsc,dfsc algorithm to sort the terms in N-gram model which effected goodly on the classification performance. The second one is proposing a new distance similarity method for N-gram model where the new method solves the problem of the difference in representation lengths among classes and documents. The third one is establishing a hybrid Center Profile Vector (CPV) classification model based on the modified N-gram and centroid classifier models. The hybrid (CPV) classification model gain a higher classification accuarcy beteer than N-gram and centroid models as the paper will show in the evaluation result.
(Keyword)
Hybrid CPV model / Text Classification / Centroid Classifier / N-gram / VSM
211.
Fuji Ren and Sohrab Golam Mohammad : Class-indexing-based term weighting for automatic text classification, Information Sciences, Vol.236, No.1, 109-125, 2013.
(Summary)
Most of the previous studies related on different term weighting emphasize on the document- indexing-based and four fundamental information elements-based approaches to address automatic text classification (ATC). In this study, we introduce class-indexingbased term-weighting approaches and judge their effects in high-dimensional and comparatively low-dimensional vector space over the TF.IDF and five other different term weighting approaches that are considered as the baseline approaches. First, we implement a class-indexing-based TF.IDF.ICF observational term weighting approach in which the inverse class frequency (ICF) is incorporated. In the experiment, we investigate the effects of TF.IDF.ICF over the Reuters-21578, 20 Newsgroups, and RCV1-v2 datasets as benchmark collections, which provide positive discrimination on rare terms in the vector space and biased against frequent terms in the text classification (TC) task. Therefore, we revised the ICF function and implemented a new inverse class space density frequency (ICSdF), and generated the TF.IDF.ICSdF method that provides a positive discrimination on infrequent and frequent terms. We present detailed evaluation of each category for the three datasets with term weighting approaches. The experimental results show that the proposed class-indexing-based TF.IDF.ICSdF term weighting approach is promising over the compared well-known baseline term weighting approaches.
(Keyword)
Text classification / Indexing / Term weighting / machine learning / Feature selection / Classifier
Changqin Quan and Fuji Ren : Finding Emotional Focus for Emotion Recognition at Sentence Level, Chinese Journal of Electronics, Vol.22, No.1, 99-103, 2013.
(Summary)
Emotion recognition at sentence level is one of the fundamental problems of textual emotion understanding. Based on the observation that sentence emotional focus can be expressed by some clauses in this sentence, this paper proposes to find the emotional focus for sentence emotion recognition. For the sake of breaking through the problems brought about by depending on emotion lexicons, we first recognize word emotions in a sentence based on Maximum entropy model. And then homogeneous Markov model is built for clause emotion recognition; After that, a strategy based on emotion selection is proposed for a sentence with multiple clauses, and genetic algorithm is used for clause selection by textual feature weighting. The experimental results show that, comparing with the baseline, there are 9.1% and 3.6% improvement respectively for two different evaluations. It is demonstrated that finding emotional focus by clause selection is able to improve the performance of sentence emotion recognition significantly.
Shin-ichi Chiba, Narimasa Watanabe, Tetsuya Tanioka, Yukie Iwasa, Kyoko Osaka, Yuko Yasuhara, Chiemi Kawanishi, Fuji Ren, Hiroshi Ogasawara and Kazushi Mifune : Use of a Dialogue System to Retrieve the Memories of Elderly Individuals with Dementia and Determine the Physiologically Effective Evaluation Indicators, International Journal of Advanced Intelligence (IJAI), Vol.4, No.1, 43-54, 2012.
214.
Yuko Yasuhara, Chiho Tamayama, Kana Kikukawa, Kyoko Osaka, Tetsuya Tanioka, Narimasa Watanabe, Shin-ichi Chiba, Masami Miyoshi, Rozzano De Castro Locsin, Fuji Ren, Shoko Fuji, Hiroshi Ogasawara and Kazushi Mifune : Required Function of the Caring Robot with Dialogue Ability for Patients with Dementia, AIA International Advanced Information Institute, Vol.4, No.1, 31-42, 2012.
215.
Seiji Tsuchiya, Motoyuki Suzuki, Fuji Ren and Hirokazu Watabe : A Novel Estimation Method of Onomatopoeic Word's Feeling based on Mora Sequence Patterns and Felling Vectors, Journal of Natural Language Processing, Vol.19, No.5, 367-379, 2012.
(Summary)
Onomatopoeic words are frequently used for expression of rich presence. These words can be understood easily for native speakers. Therefore most of onomatopoetic words are not written in a national language dictionary, or only a part of meaning is described. On the other hand, it is hard to understand a meaning of onomatopoetic words for non-native speakers. They can neither feel a meaning of an onomatopoetic word nor look it up in a dictionary. In this paper, an estimation method of feeling of an onomatopoeic word has been proposed. The feeling of the onomatopoeic word is inferred by using several features, such as morae sequence pattern of a onomatopoeic word, feeling of each mora, and so on. From the experimental results, the estimation performance of the proposed method was 0.345 (F-value). It was approximately 80% of the estimation performance given by human (F-value was 0.427). It can be said that the proposed method is useful for supporting learners of onomatopoeic words.
(Keyword)
オノマトペ / 印象推定 / モーラ系列 / 音象徴ベクトル / Onomatopoeic word / Estimation of feeling / Mora Sequence Patterns / Feeling Vectors
Ji Li and Fuji Ren : Emotion Recognition of Weblog Sentences Based on an Ensemble Algorithm of Multi-label Classification and Word Emotions, IEEJ Transactions on Electronics, Information and Systems, Vol.132, No.8, 1362-1375, 2012.
(Summary)
Weblogs have greatly changed the communication ways of mankind. Affective analysis of blog posts is found valuable for many applications such as text-to-speech synthesis or computer-assisted recommendation. Traditional emotion recognition in text based on single-label classification can not satisfy higher requirements of affective computing. In this paper, the automatic identification of sentence emotion in weblogs is modeled as a multi-label text categorization task. Experiments are carried out on 12273 blog sentences from the Chinese emotion corpus Ren_CECps with 8-dimension emotion annotation. An ensemble algorithm RAKEL is used to recognize dominant emotions from the writers perspective. Our emotion feature using detailed intensity representation for word emotions outperforms the other main features such as the word frequency feature and the traditional lexicon-based feature. In order to deal with relatively complex sentences, we integrate grammatical characteristics of punctuations, disjunctive connectives, modification relations and negation into features. It achieves 13.51% and 12.49% increases for Micro-averaged F1 and Macro-averaged F1 respectively compared to the traditional lexicon-based feature. Result shows that multiple-dimension emotion representation with grammatical features can efficiently classify sentence emotion in a multi-label problem.
Fuji Ren and Changqin Quan : Linguistic-Based Emotion Analysis and Recognition for Measuring Consumer Satisfaction - An Application of Affective computing, Information Technology and Management, Vol.13, No.4, 321-332, 2012.
(Summary)
A growing body of research suggests that affective computing has many valuable applications in enterprise systems research and e-businesses. This paper explores affective computing techniques for a vital subarea in enterprise systems consumer satisfaction measurement. We propose a linguistic-based emotion analysis and recognition method for measuring consumer satisfaction. Using an annotated emotion corpus (Ren-CECps), we first present a general evaluation of customer satisfaction by comparing the linguistic characteristics of emotional expressions of positive and negative attitudes. The associations in four negative emotions are further investigated. After that, we build a fine-grained emotion recognition system based on machine learning algorithms for measuring customer satisfaction; it can detect and recognize multiple emotions using customers' words or comments. The results indicate that blended emotion recognition is able to gain rich feedback data from customers, which can provide more appropriate follow-up for customer relationship management.
(Keyword)
Affective computing / Enterprise systems / Linguistic feature / Customer satisfaction
Yiming Tang, Fuji Ren and Yanxiang Chen : Differently implicational -universal triple I restriction method of (1, 2, 2) type, Journal of Systems Engineering and Electronics, Vol.23, No.4, 560-573, 2012.
(Summary)
From the viewpoints of both fuzzy system and fuzzy reasoning, a new fuzzy reasoning method which contains the -triple I restriction method as its particular case is proposed. The previous -triple I restriction principles are improved, and then the optimal restriction solutions of this new method are achieved, especially for seven familiar implications. As its special case, the corresponding results of -triple I restriction method are obtained and improved. Lastly, it is found by examples that this new method is more reasonable than the -triple I restriction method.
Wei Wang, Fuji Ren and Motoyuki Suzuki : A Novel Fast Fractal Image Coding Algorithm Based on Texture Feature, IEEJ Transactions on Electrical and Electronic Engineering (TEEE), Vol.7, No.5, 521-528, 2012.
(Summary)
A novel, fast fractal image coding algorithm based on the texture feature is proposed in this paper. Fractal image coding isa very promising technique for image compression. However, it has not been widely used because of the long encoding timeand high computational complexity. The most fractal image encoding time is spent in determining the approximate D-blockfrom a large D-blocks library by using the global searching method. Clustering the D-blocks library is an effective method toreduce the encoding time. First, all the D-blocks are clustered into several parts based on the new texture feature derived fromvariation function; second, for each R-block, the approximate D-blocks are searched for in the same part. In the search process,we import control parameter ; this step avoids losing the most approximate D-block for each R-block. Finally, the R-blockswhose least errors are larger than the threshold given in advance are coded by the quad tree method. We have performed asimulation with MATLABR2010a to verify the effectiveness of the proposed algorithm. The experimental results show that theproposed algorithm can be over 6 times faster than the moment-feature-based fractal image algorithm; in addition, the proposedalgorithm also improves the quality of the decoded image and increases the PSNRs average value by 2 dB. The comparisonsdemonstrate that this method is better than the fractal image coding algorithm based on statistical features
Guangwei Xu, Ming Zhu, Xin Luo, Min Wu and Fuji Ren : An unequal clustering algorithm based on energy balance for wireless sensor networks, IEEJ Transactions on Electrical and Electronic Engineering (TEEE), Vol.7, No.4, 402-407, 2012.
(Summary)
Sensor nodes in sensor networks often gather data and deliver them to the sink. However, the energy and radio range constraints force them to cooperate in transmitting the data to the destination with multihop communication. To achieve this, nodes have to be clustered and some of them are elected as the cluster head nodes to transmit the aggregated data. Unfortunately, the clustering brings extra traffic load to the cluster head nodes, particularly those closer to the sink. These nodes die faster than before as their energy is drained at a high rate. Therefore, some clustering and cluster head election algorithms were suggested to use the unequal clustering size to extend the lifetime of cluster head nodes. However, it is difficult to obtain the accurate cluster size in the process of unequal clustering. Moreover, some cluster head nodes with the overfull cluster members may die early. We propose an enhanced unequal clustering size algorithm to optimize the unequal cluster size in the different layers and balance the dissipation energy of all cluster head nodes. Simulation results show that our algorithm dissipates approximately the same energy of cluster head nodes in the different layers per round and obviously increases the number of rounds of data gathering and transmission.
Peilin Jiang, Fei Wang and Fuji Ren : Semi-Automatic Complex Emotion Categorization and Ontology Construction from Chinese Knowledge, China Communications, Vol.9, No.3, 28-37, 2012.
(Summary)
In order to recognize one's intention from the communication, both the meaning and the emotion are necessary to be interpreted correctly. But until now the study of fine-grained theory of emotion is still full of challenges. This paper analy-zes emotion category according to the statistics of Affective Word (AW) hierarchy and describes an e-motion ontology from Chinese knowledge resource semi-automatically created for human machine in-teraction. The emotion hierarchy is called complex emotion. Firstly, over 7 000 AWs have been anno-tated and their detailed explanations had been col-lected for an affective lexicon and then the con-sistent relationships are automatically parsed and a serial of emotion hierarchical structures are built up. More than 50 affective categories are extracted by a lexical clustering algorithm and about 5 000 nouns and adjectives and 2 000 verbs are categorized into the predicate hierarchy. The results have been evalu-ated to be valid by two metrics.
(Keyword)
natural language processing / affective computing / emotion ontology / emotion-top-ic variation / AW annotation
Kazuyuki Matsumoto, Kenji Kita and Fuji Ren : Emotional Vector Distance Based Sentiment Analysis of Wakamono Kotoba, China Communications, Vol.9, No.3, 87-98, 2012.
(Summary)
In this paper, we propose a method for estimating emotion in Wakamono Kotoba that were not registered in the system, by using Wakamono Kotoba example sentences as features. The proposed method applies Earth Mover's Distance(EMD) to vector of words. As a result ofthe evaluation experiment using 14,440 sentences, higher estimation accuracy is obtained by considering emotional distance between words - an approach that had not been used in the conventional research - than by using only word importance value.
XIN KANG and Fuji Ren : Predicting Complex Word Emotions and Topics through a Hierarchical Bayesian Network, China Communications, Vol.9, No.3, 99-109, 2012.
(Summary)
In this paper, we provide a Word Emotion Topic (WET) model to predict the complex word e-motion information from text, and discover the dis-tribution of emotions among different topics. A complex emotion is defined as the combination of one or more singular emotions from following 8 basic emotion categories: joy, love, expectation, surprise, anxiety, sorrow, anger and hate. We use a hierarchical Bayesian network to model the emo-tions and topics in the text. Both the complex emo-tions and topics are drawn from raw texts, without considering any complicated language features. Our experiment shows promising results of word emo-tion prediction, which outperforms the traditional parsing methods such as the Hidden Markov Model and the Conditional Random Fields(CRFs) on raw text. We also explore the topic distribution by ex-amining the emotion topic variation in an emotion topic diagram.
(Tokushima University Institutional Repository: 118261, Elsevier: Scopus)
224.
Fang Tian, Caixia Yuan and Fuji Ren : Hyponym Extraction from the Web by Bootstrapping, IEEJ Transactions on Electrical and Electronic Engineering (TEEE), Vol.7, No.1, 62-68, 2012.
(Summary)
This paper proposes an effective method to automatically extract hyponym from the Web for Chinese. The method extractshyponyms for a given hypernym through weak supervision in two stages: the first stage is submitting a hypernym and a seedhyponym as a query to Web search engine, and automatically extracting hyponyms matching with a Chinese doubly anchoredhyponymy pattern from the Web by bootstrapping. In order to reduce noise data in bootstrapping extraction, we propose a set offiltering rules to ensure matching of the proper hypernym in the extracted sentence. The second stage is ranking all the extractedcandidate hyponyms by an integrated ranking algorithm which takes into account measures both of linkage frequency betweencoordinate hyponyms and of semantic similarity between the hypernym and candidate hyponym based on co-occurrence statistics.
Yunong Wu, Kenji Kita, Fuji Ren, Kazuyuki Matsumoto and XIN KANG : Exploring the Importance of Modification Relation for Emotional Keywords Annotation and Emotion Types Recognition, International Journal of Intelligent Engineering and Systems, Vol.4, No.4, 19-26, 2011.
(Summary)
In this study, we make a scheme to explore the importance of modification relation for the emotional keywords annotation and emotion types recognition. We extract three modification features which are degree words,negative words and conjunctions from the Chinese emotion corpus named Ren-CECps. Beside word and part-of speech,three modification relations are adopted as feature in this study. We have carried out eight experiments with different feature sets for emotional keywords annotation and emotion types recognition in sentence level. Eight basic emotion types have been selected and Conditional Random Fields have been employed as the algorithm. In the part of evaluation, we demonstrate the importance of the modification features and our experiment results show the effectiveness of the modification features for improving the performance of emotional keywords annotation and emotion types recognition.
Yiming Tang, Fuji Ren and Yanxiang Chen : Reversibility of FMT-Universal Triple I Method Based on IL Operator, American Journal of Engineering and Technology Research, Vol.11, No.12, 2763-2766, 2011.
(Summary)
For the FMT-universal triple I method based on L I operator, its reversibility is investigated. Aiming at thecase that the second operator employs L I , the reversibility of FMT-universal triple I method is analyzed, where the firstoperator respectively takes seven different implication operators. It is found that the FMT-universal triple I methodseems excellent from the viewpoint of reversibility, and the second operator prefers to take L I in the universal triple Imethod.
227.
Huiwei Zhou, Xiaoyan Li, Degen Huang, Yuansheng Yang and Fuji Ren : Voting-Based Ensemble Classifiers to Detect Hedges and Their Scopes in Biomedical Texts, IEICE Transactions on Information and Systems, Vol.E94-D, No.10, 1989-1997, 2011.
(Summary)
Previous studies of pattern recognition have shown that classifiers ensemble approaches can lead to better recognition results. In this paper, we apply the voting technique for the CoNLL-2010 shared task on detecting hedge cues and their scope in biomedical texts. Six machine learning-based systems are combined through three different voting schemes. We demonstrate the effectiveness of classifiers ensemble approaches and compare the performance of three different voting schemes for hedge cue and their scope detection. Experiments on the CoNLL-2010 evaluation data show that our best system achieves an F-score of 87.49% on hedge detection task and 60.87% on scope finding task respectively, which are significantly better than those of the previous systems.
Kazuyuki Matsumoto and Fuji Ren : Estimation of word emotions based on part of speech and positional information, Computers in Human Behavior, Vol.27, No.5, 1553-1564, 2011.
(Summary)
Recently, studies on emotion recognition technology have been conducted in the fields of natural language processing, speech signal processing, image data processing, and brain wave analysis, with the goal of letting the computer understand ambiguous information such as emotion or sensibility. This paper statistically studies the features of Japanese and English emotional expressions based on an emotion annotated parallel corpus and proposes a method to estimate emotion of the emotional expressions in the sentence. The proposed method identifies the words or phrases with emotion, which we call emotional expressions, and estimates the emotion category of the emotional expressions by focusing on the three kinds of features: part of speech of emotional expression, position of emotional expression, and part of speech of the previous/next morpheme of the target emotional expression.
Kazuyuki Matsumoto, Hidemichi Sayama, Yusuke Konishi and Fuji Ren : Analysis of Wakamono Kotoba Emotion Corpus and Its Application in Emotion Estimation, International Journal of Advanced Intelligence (IJAI), Vol.3, No.1, 1-24, 2011.
(Summary)
Recently, there is a lot of research that aims to estimate emotion from text. The meaningsof linguistic expressions used in daily life vary depending on the context in which theyare used. That is to say, the information they contain presents ambiguities. Especiallythe so-called Wakamono Kotoba, Japanese language used by young people containssemantic ambiguities. Such words are usually not included in the existing dictionaries,making the meanings of these words difficult to be recognized. In this research projectwe proposed a method to estimate emotion from sentences that include Wakamono Kotobaby using statistical learning methods such as Na¨ıve Bayes method and Accumulationmethod. The existing research usually focused on learning methods using word or wordN-gram as features. However, such word-based features are insufficient to process WakamonoKotoba because Wakamono Kotoba often cannot be recognized as one semanticword by morphological analysis. In this paper we describe how we constructed the linguisticresource of Wakamono Kotoba emotion corpus to be used for emotion recognitionand introduce the features we obtained from statistical analysis. Our Wakamono Kotobaemotion corpus includes Japanese words used by young people to express emotion. Thesewere mainly gathered from Weblogs that were written by young people from their teensto their twenties.
(Keyword)
Emotion Recognition / Wakamono Kotoba / Emotion Corpus
230.
Xiao Sun, Degen Huang, Haiyu Song and Fuji Ren : Chinese New Word Identification: A Latent Discriminative Model with Global Features, Journal of Computer Science and Technology, Vol.26, No.1, 14-24, 2011.
(Summary)
Chinese new words are particularly problematic in Chinese natural language processing. With the fast development of Internet and information explosion, it is impossible to get a complete system lexicon for applications in Chinese natural language processing, as new words out of dictionaries are always being created. The procedure of new words identification and POS tagging are usually separated and the features of lexical information cannot be fully used. A latent discriminative model, which combines the strengths of Latent Dynamic Conditional Random Field (LDCRF) and semi-CRF, is proposed to detect new words together with their POS synchronously regardless of the types of new words from Chinese text without being pre-segmented. Unlike semi-CRF, in proposed latent discriminative model, LDCRF is applied to generate candidate entities, which accelerates the training speed and decreases the computational cost. The complexity of proposed hidden semi-CRF could be further adjusted by tuning the number of hidden variables and the number of candidate entities from the Nbest outputs of LDCRF model. A new-word-generating framework is proposed for model training and testing, under which the definitions and distributions of new words conform to the ones in real text. The global feature called Global Fragment Features for new word identification is adopted. We tested our model on the corpus from SIGHAN-6. Experimental results show that the proposed method is capable of detecting even low frequency new words together with their POS tags with satisfactory results. The proposed model performs competitively with the state-of-the-art models.
(Keyword)
new word identification / new words POS tagging / conditional random fields / hidden semi-CRF / global fragment features
Changqin Quan and Fuji Ren : Recognition of Word Emotion State in Sentences, IEEJ Transactions on Electrical and Electronic Engineering (TEEE), Vol.6, No.S1, 34-41, 2011.
(Summary)
Emotional word spotting has been an important step for the problems of textual emotion recognition and automatic emotionlexicon construction. To express and recognize emotion of words, especially for the words bear undirect emotions, emotionambiguity, or multiple emotions, the notion of word emotion state is proposed, which describes the state of combined basicemotions in a word. On the basis of Ren-CECps (an annotated emotion corpus) and MaxEnt (maximum entropy) modeling,we explore the effectiveness of several features and their combinations for word emotion recognition in certain contexts. Acomparative study on the performances of word emotion and word emotion state recognition is given. The experimental resultsshowed that a model using word emotion state can greatly outperform using word emotion.
(Keyword)
emotion recognition / word emotion state / Ren-CECps / MaxEnt
Kenichi Mishina, Seiji Tsuchiya, Motoyuki Suzuki and Fuji Ren : An Improvement of Example-based Emotion Estimation Using Similarity between Sentence and each Corpus, Journal of Natural Language Processing, Vol.17, No.4, 91-110, 2010.
Fuji Ren : From Cloud Computing to Language Engineering, Affective Computing and Advanced Intelligence, International Journal of Advanced Intelligence (IJAI), Vol.2, No.1, 1-14, 2010.
(Summary)
This paper discusses the definition, intension, and extension of language engineering, affective computing, and advanced intelligence, as well as the relationship among the three fields. By reporting the latest progress and future prospects, we attempt to unify language engineering and affective computing with the concept of advanced intelligence.``Cloud computing'' has recently become a very popular topic. Instead of discussing the concept and intension of cloud computing, this paper focuses on how progress in language engineering, including natural language processing and natural language understanding, will enormously aid in the achievement of cloud computing. It particularly deals with how to construct clouds, how to sweep clouds, and how to predict and exploit clouds.Another concept discussed in this paper is ``affective computing''. To a large extent, this is a breakthrough in advanced intelligence. Here, it refers to a high fusion of natural and artificial intelligence, and depends on the emotional capacity entrusted to the computer, including the capability of affective recognition and affective generation.
Changqin Quan and Fuji Ren : Sentence Emotion Analysis and Recognition Based on Emotion Words Using Ren-CECps, International Journal of Advanced Intelligence (IJAI), Vol.2, No.1, 105-117, 2010.
(Summary)
Emotion recognition on text has wide applications. In this study, we make an analysis onsentence emotion based on emotion words using Ren-CECps (a Chinese emotion corpus).Some classification methods (including C4.5 decision tree, SVM, NaiveBayes, ZEROR,and DecisionTable) have been compared. Then a supervised machine learning method(Polynomial kernel method) is proposed to recognize the eight basic emotions (Expect,Joy, Love, Surprise, Anxiety, Sorrow, Angry and Hate). Using Ren-CECps, we get theemotion lexicons for the eight basic emotions. Polynomial kernel (PK) method is usedto compute the similarities between sentences and the eight emotion lexicons. Then theexperiential knowledge derived from Ren-CECps is used to recognize whether the eightemotion categories are present in a sentence. The experiments showed promising results.
(Keyword)
Emotion recognition / Chinese emotion corpus / Classification methods / Polynomial kernel
235.
Xiao Sun, Degen Huang and Fuji Ren : Detecting New Words from Chinese Text Using Latent Semi-CRF Models, IEICE Transactions on Information and Systems, Vol.E93-D, No.6, 1386-1393, 2010.
(Summary)
Chinese new words and their part-of-speech (POS) are particularly problematic in Chinese natural language processing. With the fast development of internet and information technology, it is impossible to get a complete system dictionary for Chinese natural language processing, as new words out of the basic system dictionary are always being created. A latent semi-CRF model, which combines the strengths of LDCRF (Latent-Dynamic Conditional Random Field) and semi-CRF, is proposed to detect the new words together with their POS synchronously regardless of the types of the new words from the Chinese text without being pre-segmented. Unlike the original semi-CRF, the LDCRF is applied to generate the candidate entities for training and testing the latent semi-CRF, which accelerates the training speed and decreases the computation cost. The complexity of the latent semi-CRF could be further adjusted by tuning the number of hidden variables in LDCRF and the number of the candidate entities from the Nbest outputs of the LDCRF. A new-words-generating framework is proposed for model training and testing, under which the definitions and distributions of the new words conform to the ones existing in real text. Specific features called ``Global Fragment Information'' for new word detection and POS tagging are adopted in the model training and testing. The experimental results show that the proposed method is capable of detecting even low frequency new words together with their POS tags. The proposed model is found to be performing competitively with the state-of-the-art models presented.
(Keyword)
natural language processing / new word detection / new words POS tagging / conditional random fields / latent-dynamic CRF / semi-CRF / latent semi-CRF
Changqin Quan, Fuji Ren and Tingting He : Sentimental Classification based on Kernel Methods and Domain Semantic Orientation Dictionaries, International Journal of Innovative Computing, Information and Control, Vol.6, No.6, 2681-2690, 2010.
(Summary)
Kernel-based algorithms exploit the document information encoded in the inner-product between all pairs of document data items, avoiding explicitly the computation of the feature vector for a given input, therefore it gets considerable attention in classification tasks. In this paper, we focus our attention on the problem of sentimental classification based on three kernel methods: latent semantic kernel (LSK), polynomial kernel (PK), and Gaussian kernel (GK). It is well known that LSK has good performance in text classification, but it has relative low efficiency because of the process of the SVD decomposition, especially runs on large corpora. Our experiments demonstrate that PK has higher precision and efficiency compared with LSK and GK for the problem of sentimental classification. In particular, we compared the performances on different semantic orientation dictionaries, and found that the domain semantic orientation dictionaries could enhance the performance greatly. Also, our method could categorize the reviews with different degrees, such as 5-star, 4-star and 1-star by sorting the similarities between the reviews and the semantic orientation dictionaries. In our method, tagged corpus and certain rules are not necessary, so it is practical and has high efficiency.
Ye Liu, Zhi Teng and Fuji Ren : A Practical Sightseeing Question Answering System Based on Integrated Knowledge-Base, IEEJ Transactions on Electronics, Information and Systems, Vol.130, No.4, 580-588, 2010.
(Summary)
In this paper, a restricted domain question answering (QA) system is described. This research presents a practical sightseeing question answering system based on integrated knowledge-base. First, we use hand-crafted corpus and online resources as knowledge-base, then perform question understanding based on sightseeing place word detection and question classification. We exploit different answer extraction strategies while based on various knowledge-base (hand-crafted corpus or online resources) for answer retrieval and generation. Experimental results show the proposed method is effective for improving our former models.
(Keyword)
sightseeing question answering (QA) system / knowledge-base / hand-crafted corpus / online resources / answer generation
Ai Hakamata, Fuji Ren and Seiji Tsuchiya : Human Emotion Model Based on Discourse Sentence for Expression Generation of Conversation Agent, International Journal of Innovative Computing, Information and Control, Vol.6, No.3(B), 1537-1548, 2010.
(Summary)
There was a conversation agent on the generation method of facial expression. It is necessary for the conversation system like the human for communication. In the previous method, at first a word which could influence the feeling was defined. Facial expression was changed according to the word which influences the feeling in discourse. Whereas, facial expression could not be changed if there was not a word that was defined in the discourse. Hence, we proposed a human emotion model for the expression generation of the conversation agent. The method based on the human emotion model can solve problem of the previous method and may make a more humanity conversation agent. In this study, we put a human emotion tag to discourse of talks scenarios and model to conversation agent of human emotion. There were two kinds of methods that put the human emotion tag to discourse. We make the human emotion model by scenarios that adopts the human emotion tag. Used the human emotion model to create facial expression of conversation agent. The assessment experiment was performed by using the systems of previous method and two human emotion models, and compared the results between the three methods
Liping Mi, Xin Luo and Fuji Ren : An ERP Research on Chinese Japanese Learners' Processing of Japanese Kanji and Sentences, International Journal of Innovative Computing, Information and Control, Vol.6, No.3(B), 1491-1500, 2010.
(Summary)
This paper investigates the recognition process of Japanese Kanji and sentences for Chinese Japanese learners (CJL) and native Japanese speakers (NJS), by analyzing the event-related potential (ERP) differences between the two groups while they visually recognized Japanese Kanji and sentences The results showed that no significant differences were found between the two groups while they recognized Japanese Kanji, but significant differences were found in Japanese sentences condition, which demonstrated that the neural mechanisms of recognition process of Japanese sentences including Kana between the two groups were not identical The N400 latency of NJS appeared earlier, reflecting that NJS recognized Japanese sentences quicker than CJL did In contrast, the obvious longer N400 duration of CJL illuminated it was more difficult for CJL to recognize Japanese sentences When recognizing ambiguous sentences, CJL P600 only appeared over the right prefrontal cortex and lasted longer than that of NJS, reflecting that syntactic integration and revision of ambiguous sentences for CJL was related with the right hemisphere, and the processing load was more difficult than that of NJS. The results showed that, for CJL, the difficulty of Japanese language learning was the recognition and understanding of Japanese sentences with Kana, not Japanese Kanji
Caixia Yuan, Xiaojie Wang and Fuji Ren : Exploiting Lexical Information for Function Tag Labeling, International Journal of Innovative Computing, Information and Control, Vol.6, No.3(B), 1471-1480, 2010.
(Summary)
This paper proposes a novel approach to annotate function tags for unparsedtext. What distinguishes our work from previous attempts is that we assign function tagsdirectly basing on lexical information other than on parsed trees, thus our method isgeneral and easily portable to languages in shortage of parsing resources. In order todemonstrate the effectiveness and versatility of our method, we investigate function tagassignment for unparsed Chinese text by applying two statistical models, one is log-linearmaximum entropy model, another is maximum margin based support vector machinemodel. We show that function tag types could be determined via powerful lexical featuresand effective learning algorithms. Currently, our method achieves the best F-score of86.4 when tested on the Penn Chinese Treebank data, the highest score ever reported forChinese text.
(Keyword)
Function tags / Unparsed text / Penn treebank / Chinese language processing
Jia Ma, Motoyuki Suzuki and Fuji Ren : Spokesperson Detection Method for Autonomous Robot in Complex Communication Environment, Based on Image Processing, International Journal of Innovative Computing, Information and Control, Vol.6, No.3(B), 1515-1524, 2010.
(Summary)
Fro providing a man-machine communication system, the computer or the robot may be enabled to understand and simulate the command from human. As a pre-requisite, the system should be enabled to recognize the spokesperson, who is speaking to it However, in the realistic scene, the noise from the complex environment may cause some difficulties for the system to realize this function By imitating the performance of human, we try to use the eyes of the robot to solve the problem In this research, we propose a new spokesperson detection method based on image processing The system might find and identified all the possible spokesperson around it, by detecting and recognizing the faces in the video form the cameras on the robot It might also find out the spokesperson who is speaking to it, by judging the mouth action of possible spokesperson Using eye cameras and an omni-directional camera, we tried to realize the system and some experiments were conducted to verify the new method
(Keyword)
Spokesperson detection / Complex communication environment / Face detection / Face verification / Mouth action judgment
Changqin Quan and Fuji Ren : A blog emotion corpus for emotional expression analysis in Chinese, Computer Speech & Language, Vol.24, No.1, 726-749, 2010.
(Summary)
Weblogs are increasingly popular modes of communication and they are frequently used as mediums for emotional expression in the ever changing online world. This work uses blogs as object and data source for Chinese emotional expression analysis. First, a textual emotional expression space model is described, and based on this model, a relatively fine-grained annotation scheme is proposed for manual annotation of an emotion corpus. In document and paragraph levels, emotion category, emotion intensity, topic word and topic sentence are annotated. In sentence level, emotion category, emotion intensity, emotional keyword and phrase, degree word, negative word, conjunction, rhetoric, punctuation, objective or subjective, and emotion polarity are annotated. Then, using this corpus, we explore these linguistic expressions that indicate emotion in Chinese, and present a detailed data analysis on them, involving mixed emotions, independent emotion, emotion transfer, and analysis on words and rhetorics for emotional expression.
(Keyword)
Emotion analysis / Weblogs / Corpus annotation / Natural language processing
Fang Tian and Fuji Ren : Learning Relation Instances for Chinese Domain Ontology from the Web, IEEJ Transactions on Electrical and Electronic Engineering (TEEE), Vol.5, No.2, 188-194, 2010.
(Summary)
This paper presents a method to extract relation instances from the Internet in order to acquire knowledge that has some relations for domain ontology. We propose an ontology relation instance learning model: data sources are collected though the Web search engine and the extracted instances are constructed in Web ontology language (OWL) by Protege in Chinese. Basically, the extraction of relation instances contains syntactic patterns for filtering concepts and relevance measurement for selection of relation instances. A relevance measurement based on co-occurrence statistics is presented in this paper, which measures the semantic similarity of the measure between candidate instances and predefined domain keywords using Web search engines. In the experiment, we extract festival customs for different festival instances using relation has_custom between festival class and custom class in the Chinese festival ontology, and prove the effectiveness of our method.
Zhi Teng, Ye Liu and Fuji Ren : Create Special Domain News Collections through Summarization and Classification, IEEJ Transactions on Electrical and Electronic Engineering (TEEE), Vol.5, No.1, 56-61, 2010.
(Summary)
In this paper, we present a novel technique to create a special domain news collection system from really simple syndication (RSS) news sites through summarization and classification. The main aim of this research is to build a self-sufficient news collection system in disaster domain. In this news collection system, we used new strategies and algorithms to mine news from RSS sites, recognized and collected disaster news using automatic summarization and classification. The most striking dissimilarity between our study and previous work is that we use a novel summary approach to improve the classification performance. This paper discusses the effect of summarization and classification model on system performance. Results show that our method yields a better performance in this field.
Seiji Tsuchiya, Eriko Yoshimura, Hirokazu Watabe, Tsukasa Kawaoka and Fuji Ren : An extraction technique of place-related words based on an association mechanism, International Journal of Knowledge Engineering and Soft Data Paradigms, Vol.2, No.1, 4-14, 2010.
(Summary)
We are conducting research on the development of an intelligent robot that can converse naturally with people. Humans manipulate smooth communications by retrieving, understanding and judging several common sense concepts from conversations consciously or unconsciously. In this paper, we focus on the expression of the place from such common sense concepts. We propose a technique which is able to associate the person and thing existing in the place and the event done on the place from the word expressing the place based on an association mechanism. F-measure of the place subject and place object was 88.0% and 84.3% respectively. The average F-measure of both was 86.2%. Moreover, result of proposed technique was 37.7% better than result of a traditional technique which used case frame dictionary. These results show that proposed technique was effective and the place judgement system has achieved a judgement similar to human's sense.
Hong Zhang and Fuji Ren : Chinese POS Tagging Using Restricted Maximum Entropy Model, Chinese Journal of Electronics, Vol.19, No.1, 39-42, 2010.
(Summary)
This paper presents Chinese Part-of-speech (POS) tagging using maximum entropy technique, in which we introduce a novel gain-driven method for feature selec-tion, then we describe the restricted training method for model learning. We test our approach on the simplified Chinese corpus of Peking University China and achieve an accuracy of 97.80% and 98.60% over fine and coarse grained tag set -a significant improvement over the existing Chi-nese POS tagger.
Xiao Sun, Degen Huang and Fuji Ren : Chinese New Word Detection and POS Tagging Based on DUCRF, Information : an International Interdisciplinary Journal, Vol.12, No.6, 1349-1357, 2009.
(Summary)
In Chinese language processing, new words are particularly problematic. It is impossible to get a complete dictionary as new words call always be created. We proposed a unified Dual-chain unequal-state CRF model to detect new words together with their part-of-speech in Chinese texts regardless of the word types such as compound words, abbreviation, person names, etc. The Dual-chain Unequal-state CRF model has two stale chains with unequal number of states. The unequal state chains could model flexible hierarchical lexical information for both Chinese new word detection and POS tagging, and also integrate complex context features like the global information. The experimental results show that the proposed method is capable of detecting even low frequency new words and their parts-of-speech synchronously with satisfactory results.
(Keyword)
New word detection / Semi-CRF / Dual-chain unequal-state CRF / POS tagging
248.
Ling Xia, Zhi Teng and Fuji Ren : Question Classification for Chinese Cuisine Question Answering System, IEEJ Transactions on Electrical and Electronic Engineering (TEEE), Vol.4, No.6, 689-695, 2009.
(Summary)
This paper presents an integrated method on question classification for Chinese cuisine question answering (QA) system. First, we exploit the domain knowledge to enrich question preprocessing, then classification features are extracted by means of domain attributes and the rule-based classifier is constructed. Support vector machine (SVM) classifier is used for secondary classification to the questions which cannot be matched with rules. A prototype system based on the proposed method has been constructed and an experiment on 453 natural language questions collected from Internet has been carried out. It achieved an accuracy of 96.22%. Result shows that a small number of linguistically motivated domain features can efficiently classify questions of Chinese cuisine QA system.
Michihiro Jinnai, Satoru Tsuge, Shingo Kuroiwa, Fuji Ren and Minoru Fukumi : New Similarity Scale to Measure the Difference in Like Patterns with Noise, International Journal of Advanced Intelligence (IJAI), Vol.1, No.1, 59-88, 2009.
(Summary)
A new similarity scale called the Geometric Distance, that numerically evaluates the degree of likeness between two patterns is proposed. Traditionally, the similarity scales known as the Euclidean distance and cosine similarity have been widely used to measure likeness. Traditional methods do not perform well in the presence of noise or pattern distortions. In this paper, a new mathematical model for a similarity scale is proposed which overcomes these limitations of the earlier models, while improving the overall recognition accuracy. Experiments in speech vowel recognition were carried out under various SNR levels in a variety of noisy environments. In all cases a significant improvementin recognition accuracy is demonstrated, with the improvement most pronounced in the noisiest conditions. In fact, at a SNR of 5 dB in a subway, the recognition accuracyimproved from 65% to 75% and at 20 dB SNR from 98.4% to 99.6% over the MFCC method. Numerical modeling of simple patterns is used to demonstrate the principles behind the Geometric Distance.
Lei Yu, Xiangyang Liu, Fuji Ren and Peilin Jiang : Learning to Classify Semantic Orientation on On-line Document, International Journal of Innovative Computing, Information and Control, Vol.5, No.12, 4637-4646, 2009.
(Summary)
As a result of advance in Internet technology, automatic text sentimentclassification for a large amount of on-line documents in the form of surveys or calledreviews becomes attractive. The task of sentiment classification is to construct an effectiveclassifier with the knowledge data of vocabularies semantic meaning and the relationshipsbetween the vocabularies to determine the sentiment orientation of a document. In thispaper, one method combining HowNet knowledge base with a robust supervised sentimentclassifier is proposed. It computes semantic similarity of characteristic words and phrasesby using HowNet. Sentiment features of text are divided into characteristic words andphrases, and they adopt the positive and negative terms as features of sentiment classifier.Finally in the experiment, evaluation results show the effectiveness of our method.
(Keyword)
Sentiment classification / Text semantic orientation / Opinion mining
Peilin Jiang, Fuji Ren and Nanning Zheng : A New Approach to Data Clustering Using a Computational Visual Attention Model, International Journal of Innovative Computing, Information and Control, Vol.5, No.12, 4597-4606, 2009.
(Summary)
Cluster analysis plays an important role in many respects such as knowledge discovery, data mining and information retrieval. In this paper, we propose a newapproach inspired by the early vision system of the primate for data clustering. Humanbeings are able to locate key points that contains more important information in a complex scene. To realize this function, our approach uses a computational visual attentionmodel that selects and extracts salient areas in visual field by local difference features.Then the extracted salient areas in original visual field can be regarded as the clusters inthe data feature space. Without prior knowledge, this attention model based approach canidentify data clusters with arbitrary shapes at different scales. Finally our algorithm hasbeen tested in the evaluation experiments on the benchmark datasets to show its competitive performance.
Xiao Sun, Degen Huang and Fuji Ren : Chinese Lexical Analysis Based on Hybrid MMSM Model, International Journal of Innovative Computing, Information and Control, Vol.5, No.12, 4523-4530, 2009.
(Summary)
In this paper, we describe a scheme for Chinese word segmentation and POS tagging which integrates the character-based and word-based information in the directed graph generated by the MMSM-model. Word-level information is effective for analysis of known words, while character-level information is useful for analysis of unknown, words. A Hidden semi-CRF model is proposed for the unknown words detection. and POS tagging. The proposed Hidden semi-CRF has two state chains with unequal states which Can perform segmentation and POS tagging of unknown words simultaneously. The hybrid model was evaluated using the test data from SIGHAN-6 and achieved higher F-score than the stage-of-the-art models.
(Keyword)
Chinese morphological analysis / MMSM model / CRF / Hidden semi-CRF
Liping Mi and Fuji Ren : Facilitative Effect of the Picture Superiority Effect During Encoding and Retrieval, Science & Technology Review, Vol.27, No.20, 80-86, 2009.
(Summary)
In order to investigate the picture superiority effect, we compared the ERP between picture combined word (picture-word) and pure word (word) during encoding and retrieval. During encoding, FN400 was more negative and lasted longer for picture-word than for word. Late positive component (LPC) was more positive and distributed broadly for word compared to picture-word. During retrieval, old picture-word elicited remarkably FN400 familiarity effect and parietal old/new effect compared to old word. We suggested that simultaneous image and verbal encoding of picture-word elicited better and faster recollection compared to word during the memory test. Our results demonstrated that the picture superiority effect was related to the ability of pictures enhancing encoding and facilitating recollection.
Peilin Jiang, Lei Yu and Fuji Ren : AN CLUSTERING APPROACH FOR COMPLEX EMOTION RELATED CATEGORIZATION, ICIC Express Letters, Vol.3, No.3, 265-270, 2009.
(Summary)
An affective computing, as a branch of artificial intelligence, attracts attention as a popular growing filed with many applications insofar as information retrieval, e-learning and human computer interaction. But until now the fine-grained theory of emotion is still a challenge. In this paper, a novel method to analyze emotion related category of contemporary Chinese.
Changqin Quan, Fuji Ren and Tingting He : Word Sense Indicators: Effective Feature for Chinese Word Sense Disambiguation, Information : an International Interdisciplinary Journal, Vol.12, No.5, 1157-1164, 2009.
(Summary)
This paper presents a statistical method to extract word sense indicators as semantic disambiguators. Word sense indicators are collected based on selecting the best seeds and retrieving collocations. The similarities are comparied between sets of sense indicators for tagging sense. Our experiments showed that one or two indicators around the ambiguous word can help to recognize its sense, and sense indicators are mainly appear in ±3 context window size in Chinese.
(Keyword)
Word sense disambiguation / Word sense indicators / Mutual information
Caixia Yuan, Fuji Ren, Xiaojie Wang and Yixin Zhong : Function Labeling for Unparsed Chinese Text, IEEJ Transactions on Electronics, Information and Systems, Vol.129, No.8, 1593-1600, 2009.
(Summary)
This paper presents a work of function labeling for unparsed Chinese text. Unlike other attempts that utilize the full parse trees, we propose an effective way to recognize function labels directly based on lexical information, which is easily scalable for languages that lack sufficient parsing resources. Furthermore, we investigate a general method to iteratively simplify a sentence, thus transferring complicated sentence into structurally simple pieces. By means of a sequence learning model with hidden Markov support vector machine, we achieve the best F-measure of 87.40 on the text from Penn Chinese Treebank resources - a statistically significant improvement over the existing Chinese function labeling systems.
(Keyword)
function labeling / lexical features / sequence learning / Chinese language processing
Liping Mi, Xiangyang Liu, Fuji Ren and Hideo Araki : Characteristics of Event-related Potentials in Recognition Processes of Japanese Kanji and Sentences for Chinese Bilinguals, Journal of Physiological Anthropology, Vol.28, No.4, 191-197, 2009.
(Summary)
This paper investigates the recognition process of Japanese kanji and sentences for Chinese bilinguals and Native Japanese speakers (NJS), by analyzing the event-related potential (ERP) differences between the two groups while they visually recognized Japanese kanji and sentences. The results showed that no significant differences were found between the two groups while they recognized Japanese kanji, but significant differences were found in the Japanese sentences condition. The results demonstrated that the neural mechanisms of recognition processes of Japanese sentences including kana between the two groups were not identical. When recognizing ambiguous sentences, Chinese bilinguals' P600 only appeared over the right frontal lobe, reflecting that syntactic integration and revision of ambiguous sentences for Chinese bilinguals was related with the right hemisphere. The results showed, for Chinese bilinguals, the difficulty of Japanese language learning was the recognition and understanding of Japanese sentences with kana, not Japanese kanji. We would like to provide a scientific learning method for Chinese bilinguals to enhance their Japanese language learning efficiency from the aspect of brain science.
(Keyword)
Adult / Brain / Brain Mapping / China / Evoked Potentials / Female / Humans / Japan / Language / Male / Reading / Young Adult
Nadira Begum, Fattah Abdel Mohamed and Fuji Ren : Automatic Text Summarization Using Support Vector Machine, International Journal of Innovative Computing, Information and Control, Vol.5, No.7, 1987-1996, 2009.
(Summary)
This work investigates different text features to select the best one and proposes an approach to address automatic text summarization. This approach is a trainable summarizer, which takes into account several features, including sentence position,sentence centrality, sentence resemblance to the title, sentence inclusion of name entity,sentence inclusion of numerical data, sentence relative length, Bushy path of the sentenceand aggregated similarity for each sentence to generate summaries. First we investigatethe effect of each sentence feature on the summarization task. Then we use all featuresscore function to train Support Vector Machine (SVM) in order to construct a text summarizer model. The proposed approach performance is measured at several compressionrates (CR) on a data corpus composed of 100 English articles from the domain of politics.
(Keyword)
Automatic summarization / Support vector machine / Text features
Yun Li, Kaiyan Huang, Fuji Ren and Yixin Zhong : Wikipedia Dased Semantic Related Chinese Words Exploring and Relatedness Computing, Journal of Beijing University of Posts and Telecommunications, Vol.32, No.3, 109-112, 2009.
(Summary)
This paper introduces our way of finding semantic related Chinese word pairs from the open encyclopedia Wikipedia and analyzing the degree of semantic relations. Almost 50,000 structured documents are collected from Wikipedia pages. Then considering of hyperlinks and text overlaps etc., about 400,000 semantic related pairs are employed. We roughly measured the semantic relatedness using the position and frequency information in the documents. With comparing experiment on data sets with different degrees of semantic relations using some other classic algorithms, we analyze the reliability of our measures and other properties.
(Keyword)
Wikipedia / semantic relation / semantic relatedness
260.
Xiao Sun, Degen Huang and Fuji Ren : Chinese Lexical Analysis Based on Hidden Semi-CRF, ICIC Express Letters, An International Journal of Research and Surveys, Vol.3, No.2, 177-182, 2009.
(Summary)
In order to solve problems of the Chinese word segmentation and POS tagtaggingwhich are still existing in Chinese lexical analysis, a Hidden semi-CRF model, whichhas two chains of states with unequal number of states, is proposed for the Chinese lexicalanalysis. The Hidden semi-CRF detects the words together with their part-of-speech reregardlesswhether the words are in the system dictionary or not. A new-words-generatingframework is also built for training and testing, under which the definition and distridistributionof the new words conforms to the characteristic of the ones in real text. Theproposed framework enhances the performance of new words detecting and POS tagging,so that the overall precision of the system for Chinese lexical analysis could be furtherincreased. The experiment results show that the proposed method is capable of detectingeven low frequency new words, which in return increases the overall precision of Chineseword segmentation and POS tagging in Chinese lexical analysis.
(Keyword)
Chinese word segmentation / POS tagging / Chinese lexical analysis / Hidden semi-CRF model
Ye Yang, Peilin Jiang, Seiji Tsuchiya and Fuji Ren : Effect of Using Pragmatics Information on Question Answering System of Analects of Confucius, International Journal of Innovative Computing, Information and Control, Vol.5, No.5, 1201-1212, 2009.
(Summary)
In general, the techniques of statistical retrieval and shallow language analysis are chiefly used in question answering(QA) systems in order to improve the accuracyof answers. But these techniques are not contributing effectively to restricted domain QAsystem of classic Chinese literature such as Analects of Confucius. This QA systemrequires to extract related verses and automatically answer the query in natural Chineselanguage about Confucius thought in the Analects of Confucius. Therefore we propose a novel method to integrate the pragmatics information with classical informationretrieval technique for QA system so as to improve retrieval efficiency. In this paper,we examined the effect of the pragmatics information on QA system of the Analectsof Confucius. According to the experiments, the pragmatics information based retrievalresults are more accurate than the one without using it.
(Keyword)
Pragmatics information / Utterance interpretation / Information retrieval / Question answering system / Analects of confucius
Xiaoying Tai, Lidong Wang, Qin Chen, Fuji Ren and Kenji Kita : A New Method of Medical Image Retrieval based on Color-Texture Correlogram and GTI Model, International Journal of Information Technology & Decision Making, Vol.8, No.2, 239-248, 2009.
(Summary)
This paper presents a method for endoscopic image retrieval based on color texture correlogram and Generalized Tversky's Index (GTI) model. First we define a new image feature named color texture correlogram, which is the extension of color correlogram. The texture image extracted by texture spectrum algorithm is combined with color feature vector, and then we calculate the spatial correlation of colortexture feature vector. Similarity metric is also the key technology during domain of image retrieval, GTI model is used in medical image retrieval for similarity metric, and the technique of relevance feedback is used in the algorithm to enhance the efficiency of retrieval. Experimental results show that the method discussed in this paper is much more effective.
Yun Li, Kaiyan Huang, Fuji Ren and Yixin Zhong : Exploring Words with Semantic Relations from Chinese Wikipedia, Information : an International Interdisciplinary Journal, Vol.12, No.2, 439-449, 2009.
(Summary)
This paper introduces a way of exploring words with semantic relations from Chinese Wikipedia documents. A corpus with structured documents is generated from Chinese Wikipedia pages. Then considering of the hyperlinks, text overlaps and word frequencies, word pairs with semantic relations are explored. Words can be self clustered into groups with tight semantic relations. We roughly measure the semantic relatedness with different document based algorithms and analyze the reliability of our measures in comparing experiment.
(Keyword)
Wikipedia / Semantic Relations
264.
Kazuyuki Matsumoto, Junko Minato and Fuji Ren : Retrieval Method of Example Sentence for Intelligent English Composition Support System, Information : an International Interdisciplinary Journal, Vol.12, No.2, 377-386, 2009.
(Summary)
Recently the opportunity of reading and writing in English has been greatly increasing. We focused on supporting the Japanese to write English sentence by indicating appropriate example sentences. This paper proposed a method to retrieve such target example sentences by using domain categories and keywords or directly inputting the Japanese sentence.The result of the experiment to evaluate the precision of example sentence retrieval indicated that our technique exceeded the conventional technique by more than 20%.
265.
Ye Yang, Peilin Jiang and Fuji Ren : The Effects of a Classic Self-Learning System Utilizing Pragmatics Information and Topic, Information : an International Interdisciplinary Journal, Vol.12, No.2, 359-368, 2009.
(Summary)
The purpose of this study is to construct a new self-learning system that supports the "Analects of Confucius" learning for Chinese learner and the "Analects of Confucius" learner. We present pragmatics and topic information to the learner. There are two roles of pragmatics and topic information. The first one is to clarify the verse semantic acid the second is to make "Analects of Confucius" contents familiar to the learner. The experiments evidenced the effectiveness of the presented pragmatics and topic information that is specific to increase enthusiasm for learning and enhance understanding of the contents.
(Keyword)
self-learning system / pragmatics information / enthusiasm for learning / understanding of the contents / linguistic education
266.
Ye Wu and Fuji Ren : Emotion Recognition Based on Negative Words and Pattern Matching For Chinese Negative Sentences, Information : an International Interdisciplinary Journal, Vol.12, No.2, 259-267, 2009.
(Summary)
Abstract Recently, emotion recognition with computer has attracted a great deal of attention to researchers for its broad applications. Emotion estimation from textual input has also become active as natural language processing (NLP) technology develops. However, when it comes to negative sentences in Chinese, the original emotion estimation may be reversed which makes obtaining correct recognition results difficult if we do not consider the effect of negative words.
267.
Hong Zhang and Fuji Ren : Negative Expression Translation for Japanese and Chinese Machine Translation System, Information : an International Interdisciplinary Journal, Vol.12, No.2, 247-257, 2009.
(Summary)
There are many differences between Chinese and Japanese in history, culture, life, manners and customs, therefore it is natural that there are particular words and special usage of words in two languages. It is necessary to correctly master the meaning of these particular words and then translate accurately between two languages. Because of the complex corresponding relationship in Chinese and Japanese, when building Japanese-Chinese machine translation, it is easy to introduce vagueness. Among the mistakenly translations in existing commercial translation software, most of them are caused by the negative expression in the sentence. In this paper, through analyzing the negative expression ways in Chinese and Japanese languages, we investigated the translation of Japanese-Chinese negative sentences by using the selection rules of Chinese negative words and position rules. The basic Japanese negative expression naide/(sic) has the same meaning with mei/(sic) and bie/(sic) in Chinese. When translating nakute/(sic) into Chinese, according to its negative meaning and grammatical rule, it may be translated into bu/(sic). In the current research, we discussed the general rules which are abstracted from typical examples including naide/(sic) and nakute/(sic).
(Keyword)
Negative expression / Japanese and Chinese / Machine Translation
Ye Yang, Seiji Tsuchiya and Fuji Ren : Construction of "Analects of Confucius" Knowledge Base Including Pragmatics Information, IEEJ Transactions on Electronics, Information and Systems, Vol.129, No.1, 133-139, 2009.
(Summary)
In this paper, we propose an approach of constructing the knowledge base for ``Analects of Confucius'', which aims to help the correct understanding of ``Analects of Confucius''. The content of ``Analects of Confucius'' has the characteristics that it has not been categorized by topics and always has comprehensive meanings. Thus it is necessary to build an framework to manage and build the knowledge base for it. The construction of knowledge base in the past work has been focusing on the research of the words and the shallow meaning and explicitly communicated meaning of the passages which can not be used for deeper meaning and implicitly communicated meaning. The present paper sets up an categorization system for ``Analects of Confucius'', then based on it, the knowledge base is built by using pragmatics information with reference to utterance interpretation method in pragmatics. The question answering system adopts the knowledge base of ``Analects of Confucius'' and gives an assessment of it. The results show that the combined answer with pragmatics information and modern text interpretation outperforms the answers which are extracted only from modern text by 32.2%.
(Keyword)
Analects of Confucius / Pragmatics / Pragmatics Information / Knowledge Base / Analects of Confucius / Pragmatics / Pragmatics Information / Knowledge Base
Yu Zhang and Fuji Ren : Statistic Information and Grammatical Features Based Emotion Recognition For Chinese Split Phrases, Information : an International Interdisciplinary Journal, Vol.12, No.1, 183-191, 2009.
(Summary)
Emotion recognition in context is a new research issue with great significance. It can help people acquire the useful information from reviews or comments in the Internet. In Chinese context, there is a syntactic construction called split phrase that always carries emotional information, which cannot be recognized accurately by general lexical analysis. In order to find an effective way for recognizing this construction, we study the classification and calculation rules of split phrases in Chinese. In this paper, an emotion recognizing method based on the statistic information and grammatical feature is proposed and evaluated. The comparing experimental results show that the emotion recognition rate is improved efficiently.
(Keyword)
Emotion Recognition / Chinese Split Phrase / Emotion Classification
270.
Jiajun Yan, David B. Bracewell, Fuji Ren and Shingo Kuroiwa : Integration of Multiple Classifiers for Chinese Semantic Dependency Analysis, Electronic Notes in Theoretical Computer Science, Vol.225, No.1, 457-468, 2009.
(Summary)
Semantic Dependency Analysis (SDA) has extensive applications in Natural Language Processing (NLP). In this paper, an integration of multiple classifiers is presented for SDA of Chinese. A Naive Bayesian Classifier, a Decision Tree and a Maximum Entropy classifier are used in a majority wins voting scheme. A portion of the Penn Chinese Treebank was manually annotated with semantic dependency structure. Then each of the three classifiers was trained on the same training data. All three of the classifiers were used to produce candidate relations for test data and the candidate relation that had the majority vote was chosen. The proposed approach achieved an accuracy of 86% in experimentation, which shows that the proposed approach is a promising one for semantic dependency analysis of Chinese.
Manabu Sasayama, Fuji Ren and Shingo Kuroiwa : Automatic Extraction of Super-Function From Bilingual Corpus, Electronic Notes in Theoretical Computer Science, Vol.225, No.1, 329-340, 2009.
(Summary)
Extraction of a large Super-Function (SF) is one of the most important factor in realizing SF based machine translation. This paper presents a method to automatically extract SF from a Japanese-English bilingual corpus. The extraction process matches Japanese noun and English noun in each bilingual sentence in a bilingual corpus using a bilingual dictionary. The experimental results show that this method performs very well in automatically extracting SF for machine translation. Then, we discuss a problem of SF based machine translation from the result of the evaluation experiment using extracted SF.
Fuji Ren and Bracewell B. David : Advanced Information Retrieval, Electronic Notes in Theoretical Computer Science, Vol.225, No.1, 303-317, 2009.
(Summary)
In this paper we explore some of the most important areas of advanced information retrieval. In particular, we look at cross-lingual information retrieval, multimedia information retrieval and semantic-based information retrieval. Cross-lingual information retrieval deals with asking questions in one language and retrieving documents in one or more different languages. With an increasingly globalized economy, the ability to find information in other languages is becoming a necessity. Multimedia information retrieval deals with finding media other than text, i.e. music and pictures. With the explosion of digital media that is available on the Internet and present on users' computers techniques for quickly and accurately finding desired media is important. Semantic based information retrieval goes beyond classical information retrieval and uses semantic information to understand the documents and queries in order to aid retrieval. Semantic based information retrieval goes beyond standard surface information by using the concepts represented in documents and queries to improve retrieval performance.
Bracewell B. David, Jiajun Yan, Fuji Ren and Shingo Kuroiwa : Category Classification and Topic Discovery of Japanese and English News Articles, Electronic Notes in Theoretical Computer Science, Vol.225, No.1, 51-65, 2009.
(Summary)
This paper presents algorithms for topic analysis of news articles. Topic analysis entails category classification and topic discovery and classification. Dealing with news has special requirements that standard classification approaches typically cannot handle. The algorithms proposed in this paper are able to do online training for both category and topic classification as well as discover new topics as they arise. Both algorithms are based on a keyword extraction algorithm that is applicable to any language that has basic morphological analysis tools. As such, both the category classification and topic discovery and classification algorithms can be easily used by multiple languages. Through experimentation the algorithms are shown to have high precision and recall in tests on English and Japanese.
Fuji Ren : Affective Information Processing and Recognizing Human Emotion, Electronic Notes in Theoretical Computer Science, Vol.225, 39-50, 2009.
(Summary)
Information recognition and extraction of human emotions are necessary for machines to communicate smoothly with humans and to realize emotion communications. We focus on human psychological characteristics to develop general-purpose agents that can recognize human emotion and create machine emotion. We comprehensively analyze brain waves, voice sounds and picture images that represent information included in emotion elements of phonation, facial expressions, and speech usage. We analyze and estimate many statistical data based on the latest achievements of brain science and psychology in order to derive transition networks for human psychological states. We establish a speaker word model for researching computer simulation of psychological change and emotional presentation, developing emotion interface, and establishing theoretic structure and realization method of emotion communication. A new approach for recognizing human emotion based on Mental State Transition Network will be described and one emotion estimation method based on sentence pattern of emotion occurrence events will be discussed, and some new results of the project will be given.
(Keyword)
recognizing human emotion / emotion communication / Information recognition / recognize human emotion / create machine emotion
Kazuyuki Matsumoto, Tetsuya Tanioka, Kyoko Osaka, Kawamura Ai, Shu-ichi Ueno, Fuji Ren, Takasaka Yoichiro, Barnard Alan, Rozzano De Castro Locsin and Omori Mitsuko : Developing the Method of Server Controlled Outcomes Management and Variance Analysis, Electronic Notes in Theoretical Computer Science, Vol.225, 221-237, 2009.
(Summary)
This paper is to describe a way to develop the Psychoms process for mental health patient management and variance analysis system has been developed using artificial intelligence. Although it is agreed that there is a need for clinical pathway variance analysis, methods for creating a system are less well defined. The procedure and systematic process described aims to improve patients' quality of life through consistent and timely care. Ultimately, its potential influence is to assist in the improvement of quality health care services. This paper illustrates a method of outcomes management and variance analysis as the prospective development of future research.
(Keyword)
Outcome Management / variance analysis / nursing administration / team care / artificial intelligence
Fattah Abdel Mohamed and Fuji Ren : GA, MR, FFNN, PNN & GMM based Models for Automatic Text Summarization, Computer Speech & Language, Vol.23, No.1, 126-144, 2009.
(Summary)
This work proposes an approach to address the problem of improving content selection in automatic text summarization by using some statistical tools. This approach is a trainable summarizer, which takes into account several features, including sentence position, positive keyword, negative keyword, sentence centrality, sentence resemblance to the title, sentence inclusion of name entity, sentence inclusion of numerical data, sentence relative length, Bushy path of the sentence and aggregated similarity for each sentence to generate summaries. First, we investigate the effect of each sentence feature on the summarization task. Then we use all features in combination to train genetic algorithm (GA) and mathematical regression (MR) models to obtain a suitable combination of feature weights. Moreover, we use all feature parameters to train feed forward neural network (FFNN), probabilistic neural network (PNN) and Gaussian mixture model (GMM) in order to construct a text summarizer for each model. Furthermore, we use trained models by one language to test summarization performance in the other language. The proposed approach performance is measured at several compression rates on a data corpus composed of 100 Arabic political articles and 100 English religious articles. The results of the proposed approach are promising, especially the GMM approach.
Kyoko Osaka, Seiji Tsuchiya, Fuji Ren, Shingo Kuroiwa, Tetsuya Tanioka and Rozzano De Castro Locsin : The Technique of Emotion Recognition Based on Electroencephalogram, Information : an International Interdisciplinary Journal, Vol.11, No.1, 55-68, 2008.
(Summary)
We aim to develop the mechanism that is possible to sympathize with man, and we target at the thing to read feelings which man feels from the brain wave. This time, as an initial stage of the research how the subject is able to make judgments to be impressed from the brain wave is verified. Concretely, it is investigated by using electroencephalograph (EEG) that the brain is an active state when the subject own declaring to have been impressed. Three kinds of evaluation method are used for this research. One method is statistically evaluated based on the strength of potential. Other method is evaluated objective based on the place which brain waves activate. Another method is evaluated by comparing a subject's subjectivity with change of EEG. Subjects are two persons and a small number this time, and since those attributes are partial, a question remains in the justification of a result. However, it is also the fact which becomes clear from this result that a subject's impression condition can fully be judged from the activity state of brain waves.
278.
Manabu Sasayama, Shingo Kuroiwa and Fuji Ren : Extracting Date/Time Expressions in Super-Function Based Japanese-English Machine Translation, IEEJ Transactions on Electronics, Information and Systems, Vol.128, No.8, 1342-1350, 2008.
(Summary)
Super-Function Based Machine Translation(SFBMT) which is a type of Example-Based Machine Translation has a feature which makes it possible to expand the coverage of examples by changing nouns into variables, however, there were problems extracting entire date/time expressions containing parts-of-speech other than nouns, because only nouns/numbers were changed into variables. We describe a method for extracting date/time expressions for SFBMT. SFBMT uses noun determination rules to extract nouns and a bilingual dictionary to obtain correspondence of the extracted nouns between the source and the target languages. In this method, we add a rule to extract date/time expressions and then extract date/time expressions from a Japanese-English bilingual corpus. The evaluation results shows that the precision of this method for Japanese sentences is 96.7%, with a recall of 98.2% and the precision for English sentences is 94.7%, with a recall of 92.7%.
(Keyword)
date / time expression / Super-Function / machine translation / time expression / machine translation
Fattah Abdel Mohamed and Fuji Ren : Sentence Alignment based on the Text Length between Punctuation Marks, Information : an International Interdisciplinary Journal, Vol.11, No.4, 445-465, 2008.
(Summary)
Sentence alignment based on the text length between punctuation marks | sentence alignment; English/Arabic parallel corpus; parallel corpora; machine translation | Parallel corpora have become an essential resource for work in multi lingual natural language processing. Sentence aligned parallel corpora are more efficient than non-aligned parallel corpora for cross language information retrieval and machine translation applications. In this paper, we present a new approach to align sentences in bilingual parallel corpora based on the text character length between successive punctuation marks. A probabilistic score is assigned to each proposed correspondence of texts, based on the scaled difference of lengths of the two texts (in characters) and the variance of this difference. Using this score, the time required for punctuation marks matching decreased and the sentence alignment accuracy increased. Using this new approach, we could achieve an error reduction of 26.5% over length based approach when applied on English-Arabic parallel documents. The sentence alignment execution time decreased to 17% of the total time required for the combined model which uses length based approach and punctuation approach combined together. Moreover, the proposed approach result outperforms Melamed and Moore's approach results.
(Keyword)
sentence alignment / English / Arabic parallel corpus / parallel corpora / machine translation / WEB
280.
Junko Minato, Kazuyuki Matsumoto, Fuji Ren, Seiji Tsuchiya and Shingo Kuroiwa : Evaluation of Emotion Estimation Methods Based on Statistic Features of Emotion Tagged Corpus, International Journal of Innovative Computing, Information and Control, Vol.4, No.8, 1931-1941, 2008.
(Summary)
Development of information technology has been increasing the chance of interaction between human and computer. Computer used to be a tool and its development had been mainly emphasized on the speed of information processing. However, by understanding not only semantic information but also ambiguous information such as emotion, computer can be a versatile partner for human that can detect emotional state of a person and can response appropriately the emotion. As a primary study of text-based emotion estimation by computer we focused on statistically analyzing the relationship between word emotion and sentence emotion based on the originally created emotion tagged corpus then propose two emotion estimation methods according to the obtained statistic features. The proposed methods are evaluated using a prototype system.
(Keyword)
Corpus analysis / Affective computing / Statistic analysis
Fattah Abdel Mohamed and Fuji Ren : English-Arabic Proper Noun Transliteration Pairs Creation, Journal of the American Society for Information Science and Technology, Vol.59, No.10, 1675-1687, 2008.
(Summary)
Proper nouns may be considered the most important query words in information retrieval. If the two languages use the same alphabet, the same proper nouns can be found in either language. However, if the two languages use different alphabets, the names must be transliterated. Short vowels are not usually marked on Arabic words in almost all Arabic documents (except very important documents like the Muslim and Christian holy books). Moreover, most Arabic words have a syllable consisting of a consonant-vowel combination (CV), which means that most Arabic words contain a short or long vowel between two successive consonant letters. That makes it difficult to create English-Arabic transliteration pairs, since some English letters may not be matched with any romanized Arabic letter. In the present study, we present different approaches for extraction of transliteration proper-noun pairs from parallel corpora based on different similarity measures between the English and romanized Arabic proper nouns under consideration. The strength of our new system is that it works well for low-frequency proper noun pairs. We evaluate the new approaches presented using two different English-Arabic parallel corpora. Most of our results outperform previously published results in terms of precision, recall, and F-Measure.
(Keyword)
INFORMATION-RETRIEVAL / PARALLEL CORPORA / EXTRACTION / ALIGNMENT / WORDS
Manabu Sasayama, Fuji Ren and Shingo Kuroiwa : Automatic Super-function Extraction for Translation of Spoken Dialogue, International Journal of Innovative Computing, Information and Control, Vol.4, No.6, 1371-1382, 2008.
(Summary)
Extraction of a large number of Super-Function (SF) is the most importantfactor in realizing SF based machine translation. This paper presents a method for automatic extraction of SF from a Japanese-English bilingual corpus. The extraction processuses a bilingual dictionary to match Japanese and English nouns in each sentence pair.The experimental results using a Japanese-English bilingual corpus show that this methodperforms very well in automatically extracting SF for machine translation. In addition,we evaluate the extracted SF in SF based machine translation.
Liying Mi, Xin Luo and Fuji Ren : Chinese-Japanese Translation of Causative Sentences Using Super-function Based Machine Translation System, International Journal of Innovative Computing, Information and Control, Vol.4, No.4, 915-926, 2008.
(Summary)
Causative sentences in Japanese are a matter of affixation. In particular,the causative form takes the shape of a suffix. In Chinese, the causative form constitutes an independent word. In our previous studies on Super-Function Based MachineTranslation (SFBMT), we have found that causative sentences are very frequently usedand difficult to translate correctly, the over use of causative sentences can be dangerousas it may introduce ambiguity in the translation. In this paper, we discuss the challenges in handling Japanese causative sentences in an SFBMT system; we present ashallow method for translating causative sentences by using some fixed rules and SuperFunctions (SF). In the present research, sufficient Chinese-Japanese causative sentencepatterns have been employed as a language-database for experiments, which proves thesuggested method can effectively improve translation quality within the range under discussion.
Bracewell B. David, Jiajun Yan and Fuji Ren : Single Document Keyword Extraction For Internet News Articles, International Journal of Innovative Computing, Information and Control, Vol.4, No.4, 905-914, 2008.
(Summary)
Keywords are a fundamental part of information retrieval (IR) and as suchthey have been studied extensively. They are used for everything from searching to describing a document. A Keyword extraction algorithm can be defined as a combinationof a keyword representation and a selection/weighting scheme. The most common selection/weighting schemes are based on collection statistics or using supervised machinelearning algorithms. In these cases, keywords can, typically, only be extracted from documents that belong to a collection or using a large amount of annotated training data. Theimportance of extracting keywords without a document collection has been gradually increasing due to the Internet. In this paper, a keyword extraction algorithm designed withnews in mind that requires neither a document collection or training data is presented.It uses noun phrases as its keyword representation and takes in document statistics toderive its weighting scheme. Through experimentation it is shown that the quality of thekeywords extracted from the proposed algorithm are better than standard algorithms forboth information retrieval and humans.
(Keyword)
Keyword extraction / Indexing / news / information retrieval system / Natural language processing
Jiajun Yan, David B. Bracewell, Fuji Ren and Shingo Kuroiwa : The Creation of a Chinese Emotion Ontology Based on HowNet, Engineering Letters, Vol.16, No.1, 166-171, 2008.
(Summary)
Full comprehension of language comes about by understanding the meaning and the emotion behind the communication. Understanding the meaning of language is the goal of natural language processing and research on semantic analysis. Understanding emotion is one of the goals of affective computing. The two areas of artificial intelligence have recently come together for understanding emotion in text. In order to help in this pursuit, this paper describes a Chinese emotion ontology based on HowNet and its construction. The ontology should go a long way in helping to understand, classifiy, and recognize emotion in Chinese. The ontology created in this paper is made up of Chinese emotion predicates that can help in understanding the emotion of the actors in sentences. The ontology was semi-automatically created using a simple three step process. The final ontology has just under 5,500 verb predicates covering 113 different emotion categories.
(Keyword)
Natural Language Processing / Emotion Ontology / Affective Computing / Chinese
286.
David B. Bracewell, Fuji Ren and Shingo Kuroiwa : A Low Cost Machine Translation Method for Cross-Lingual Information Retrieval, Engineering Letters, Vol.16, No.1, 160-165, 2008.
(Summary)
In one form or another language translation is a necessary part of cross-lingual information retrieval systems. Often times this is accomplished using machine translation systems. However, machine translation systems offer low quality for their high costs. This paper proposes a machine translation method that is low cost while improving translation quality. This is done by utilizing multiple web based translation services to negate the high cost of translation. A best translation is chosen from the candidates using either consensus translation selection or statistical analysis. Which to use is determined by a heuristic rule that takes into account that most web based translation services are of similar quality and that machine translation still produces relatively poor results. By choosing the best translation the method is able to increase translation quality over the base systems, which is verified by the experimentation.
(Keyword)
Natural Language Processing / Statistical Analysis / Machine Translation / Cross-Lingual Information Retrieval
287.
Junko Minato, Bracewell B. David, Fuji Ren and Shingo Kuroiwa : Japanese Emotion Corpus Analysis and its Use for Automatic Emotion Word Identification, Engineering Letters, Vol.16, No.1, 172-177, 2008.
(Summary)
In this paper, the creation of a Japanese emotion cor- pus and its use in automatic emotion word identifcation are examined. The corpus was created by manually tagging words in just under 1,200 dialog sentences with emotion. Using the tagged corpus, statistical anal- ysis was performed to determine the characteristics of emotional expression in Japanese dialog. This type of analysis should prove benefcial for understanding how emotion is expressed and how to identify, classify, etc. emotion in Japanese. To test this theory an automatic emotion word identifcation system was built using machine learning based classifers with features taken from the statistical analysis. In total, four diferent classifers were trained and compared to a baseline dictionary approach. It was found that classifer based identifcation was able to signifcantly increase recall.
(Keyword)
Natural Language Processing / Statistical Analysis / Affective Computing / Corpus Creation
288.
Fattah Abdel Mohamed and Fuji Ren : Automatic Text Summarization, International Journal of Computer Science, Vol.3, No.1, 25-28, 2008.
(Summary)
This work proposes an approach to address automatic text summarization. This approach is a trainable summarizer, which takes into account several features, including sentence position, positive keyword, negative keyword, sentence centrality, sentence resemblance to the title, sentence inclusion of name entity, sentence inclusion of numerical data, sentence relative length, Bushy path of the sentence and aggregated similarity for each sentence to generate summaries. First we investigate the effect of each sentence feature on the summarization task. Then we use all features score function to train genetic algorithm (GA) and mathematical regression (MR) models to obtain a suitable combination of feature weights. The proposed approach performance is measured at several compression rates on a data corpus composed of 100 English religious articles. The results of the proposed approach are promising.
(Keyword)
Automatic Summarization / Genetic Algorithm / Mathematical Regression / Text Features
289.
Kyoko Osaka, Tetsuya Tanioka, Shuichi Ueno, Chiemi Kawanishi, Toshiko Tada, Shingo Kuroiwa and Fuji Ren : Empathetic Understanding as Caring in Nursing Using Electroencephalographic Data as Evidence, International Journal for Human Caring, Vol.12, No.1, 7-16, 2008.
(Summary)
We presume that the measurement of electroencephalographic (EEG) changes, those activities that are considered physiological indicators, enables an objective understanding of changes in emotions of those who have difficulty in expressing these through facial expression or physical action. Generally, EEG is used in the hospital to examine encephalopathy and brain disorder. Using an electroencephalograph device to acquire digital data we propose a method to objectively capture changes in the recognition state of people from changes in EEG activities (action potential), and a way to apply it into a clinical situation.
290.
Dapeng Yin, Min Shao, Fuji Ren and Shingo Kuroiwa : Improving Parsing of BA Sentences for Machine Translation, IEEJ Transactions on Electrical and Electronic Engineering (TEEE), Vol.3, No.1, 106-112, 2008.
(Summary)
The research on Chinese-Japanese machine translation has been lasting for many years, and now this research field is increasingly thoroughly refined. In practical machine translation system, the processing of a simple and short Chinese sentence has somewhat good results. However, the translation of complex long Chinese sentence still has difficulties. For example, these systems are still unable to solve the translation problem of complex BA sentences. In this article a new method of parsing of BA sentence for machine translation based on valency theory is proposed. A BA sentence is one that has a prepositional word BA. The structural character of a BA sentence is that the original verb is behind the object word. The object word after the BA preposition is used as an adverbial modifier of an active word. First, a large number of grammar items from Chinese grammar books are collected, and some elementary judgment rules are set by classifying and including the collected grammar items. Then, these judgment rules are put into use in actual Chinese language and are modified by checking their results instantly. Rules are checked and modified by using the statistical information from an actual corpus. Then, a five-segment model used for BA sentence translation is brought forward after the above mentioned analysis. Finally, we applied this proposed model into our developed machine translation system and evaluated the experimental results. It achieved a 91.3% rate of accuracy and the satisfying result verified effectiveness of our five-segment model for BA sentence translation.
(Keyword)
machine translation / five-segment model / Chinese / Japanese
Peilin Jiang, Hua Xiang and Fuji Ren : The Framework of Mental State Transition Analysis, Lecture Notes in Computer Science, Vol.4827, 1046-1055, 2007.
(Summary)
The Human Computer Interaction (HCI) Technology has emerged in the different fields in applications in computer vision and recognition systems such as virtual environment, video games, e-business and multimedia management. In this paper we propose a framework of designing the Mental State Transition (MST) of a human being or virtual character. The expressions of human emotion can be easily remarked by facial expressions, gestures, sound and other visual characteristics. But the potential MST modeling in affective data are always hidden actually. We analysis the framework of MST and employ DBNs to construct the MST networks and finally the experiment has been implemented to derive the ground truth of the data and verify the effectiveness.
(Keyword)
Mental State Transition / HCI / Psychological Experiment / Virtual Character
Kazuyuki Matsumoto, Fuji Ren, Shingo Kuroiwa and Seiji Tsuchiya : Emotion Estimation Algorithm Based on Interpersonal Emotion Included in Emotional Dialogue Sentences, Lecture Notes in Computer Science, Vol.4827, 1035-1045, 2007.
(Summary)
Emotion recognition aims to make computer understand ambiguous information of human emotion. Recently, research of emotion recognition is actively progressing in various fields such as natural language processing, speech signal processing, image data processing or brain wave analysis. We propose a method to recognize emotion in dialogue text by using originally created Emotion Word Dictionary. The words in the dictionary are weighted according to the occurrence rates in the existing emotion expression dictionary. We also propose a method to judge the object of emotion and emotion expressivity in dialogue sentences. The experiment using 1,190 sentences proved about 80% accuracy.
Peilin Jiang, Ran Li, Fuji Ren, Shingo Kuroiwa and Nanning Zheng : Color Features Based Speaking Detection with Hidden Markov Model in Video Sequences, Research in Computing Science, Vol.32, 374-381, 2007.
(Summary)
The Human Computer Interface Technology has faced challenges of understanding user's mind actively. In the ¯rst, the speak detection is a primary technique in applications of human computer interface(HCI) and other applications like surveillance system, video conferenceand multimedia data base management in computer vision and speechrecognition. This paper describes a novel method to detect speaker witha probabilistic model of behavior of speaking. After human face recognition, the especial components under the nonlinear transformation incolor space of lip represent the speci¯c mouth region and then combine the groups of coherent motions . Next the simple movements in themouth region are modeled by hidden Markov models. The experimentalresults demonstrate that the model representing speaking is e±ciencyand successful in applying to driver video surveillance system.
294.
Yu Zhang, Zhuoming Li, Fuji Ren and Shingo Kuroiwa : A Construction of Emotion Thesaurus Basing on Chinese Character and Empirical Knowledge, Research in Computing Science, Vol.32, 330-340, 2007.
(Summary)
There have been some studies about spoken natural language dialog, and most of them have successfully been developed within the specified domains. However, current human-computer interfaces only get the data to process their programs. Aiming at developing an affective dialog system, we have been exploring how to incorporate emotional aspects of dialog into existing dialog processing techniques. As a preliminary step toward this goal, we work on making a Chinese emotion classification model which is used to recognize the main affective attribute from a sentence or a text. Finally we have done experiments to evaluate our model.
295.
Kazuma Hara, Shingo Kuroiwa, Kouji Tanaka, Satoru Tsuge, Fuji Ren, Masami Shishibori and Kenji Kita : Acoustic Model Adaptation Using Speech Synthesis for Codec Speech Recognition, The Transactions of the Institute of Electronics, Information and Communication Engineers D, Vol.J90-D, No.9, 2541-2549, 2007.
Xin Luo, Masami Shishibori, Fuji Ren and Kenji Kita : Incorporate feature space transformation to content-based image retrieval with relevance feedback, International Journal of Innovative Computing, Information and Control, Vol.3, No.5, 1237-1250, 2007.
(Summary)
In recent years, the employment of feedback information to improve image retrieval precision has become a hot subject in research field. But in the traditional relevance feedback methods, both relevant and non-relevant assigned information were required for the retrieval system. For the sake of practicality and convenience, this paper assumes that users only need to label the images from the previous query as relevant, which generates a new vector as feedback information. Through the feature space transformation, it is an adjustment in the spatial resolution of the feature space. The spatial resolution around relevant samples is contracted. Meanwhile, the analysis of the user's intension together with relevant forecast of the interested objects in the database make it posiible for the less similar vectors to get closer to the query vector and thus increasing query precision. Compared with the traditional relevance feedback approach, our method is shown to obviously improve the retrieval feedback performance.
Shingo Kuroiwa, Satoru Tsuge, Masahiko Kita and Fuji Ren : Speaker Identification Method Using Earth Mover's Distance for CCC Speaker Recognition Evaluation 2006, International Journal of Computational Linguistics & Chinese Language Processing, Vol.12, No.3, 239-254, 2007.
(Summary)
In this paper, we present a non-parametric speaker identification method using Earth Movers Distance (EMD) designed for text-indepedent speaker identification and its evaluation results for CCC Speaker Recognition Evaluation 2006, organized by the Chinese Corpus Consortium (CCC) for the th International Symposium on Chinese Spoken Language Processing (ISCSLP 2006). EMD based speaker identification (EMD-IR) was originally designed to be applied to a distributed speaker identification system, in which the feature vectors are compressed by vector quantization at a terminal and sent to a server that executes a pattern matching process. In this structure, we had to train speaker models using quantized data, then we utilized a non-parametric speaker model and EMD. From the experimental results on a Japanese speech corpus, EMD-IR showed higher robustness to the quantized data than the conventional GMM technique. Moreover, it achieved higher accuracy than GMM even if the data was not quantized. Hence, we have taken the challenge of CCC Speaker Recognition Evaluation 2006 using EMD-IR. Since the identification tasks defined in the evaluation were on an open-set basis, we introduce a new speaker verification module. Evaluation results show that EMD-IR achieves 99.3 % Identification Correctness Rate in a closed-channel speaker identification task.
(Keyword)
Speaker Identification / Earth Movers Distance / Non-Parametric / Vector Quantization / Chinese Speech Corpus
298.
Zhi Teng, Fuji Ren and Shingo Kuroiwa : An Emotion Recognition Conversation System Based on Knowledge Database Automatic Architecture, Communications in Computer and Information Science, Vol.2, No.17, 722-731, 2007.
(Summary)
As conversation system it must converse with various users and have to talk about various conversations, so we have to prepare large numbers of information for conversation knowledge database, but it is so especially difficult for us. The wealth of information on the web is full and quick so those make it an attractive resource for seeking quick information to simple. Recognizing a human's emotional state can be helpful in various contexts. The most promising one is probably the man-machine interaction, the communication between an assisting robot in the household and its human user. In this paper we present an experiment named an emotion recognition conversation system based on knowledge database automatic architecture. In this exploration there are two parts different than the generic conversation system. First, our conversation system can automatic construct of knowledge database base on the Really Simple Syndication parse. Second, this system can recognize emotion from the conversation contents. The experiment showed that this method could achieve better results in practice.
(Keyword)
Conversation System / Conversation Knowledge Database / Emotion Recognition
Jiajun Yan, Bracewell B. David, Shingo Kuroiwa and Fuji Ren : Chinese semantic dependency analysis: Construction of a treebank and its use in classification, ACM Transactions on Speech and Language Processing, Vol.4, No.2, 1-20, 2007.
(Summary)
Semantic analysis is a standard tool in the Natural Language Processing (NLP) toolbox with widespread applications. In this article, we look at tagging part of the Penn Chinese Treebank with semantic dependency. Then we take this tagged data to train a maximum entropy classifier to label the semantic relations between headwords and dependents to perform semantic analysis on Chinese sentences. The classifier was able to achieve an accuracy of over 84%. We then analyze the errors in classification to determine the problems and possible solutions for this type of semantic analysis.
(Keyword)
Chinese / Natural language processing / maximum entropy classifiation / semantic dependency analysis
Fattah Abdel Mohamed, Bracewell B. David, Fuji Ren and Shingo Kuroiwa : Sentence alignment using P-NNT and GMM, Computer Speech & Language, Vol.21, No.4, 594-608, 2007.
(Summary)
Parallel corpora have become an essential resource for work in multilingual natural language processing. However, sentence aligned parallel corpora are more efficient than non-aligned parallel corpora for cross-language information retrieval and machine translation applications. In this paper, we present two new approaches to align EnglishArabic sentences in bilingual parallel corpora based on probabilistic neural network (P-NNT) and Gaussian mixture model (GMM) classifiers. A feature vector is extracted from the text pair under consideration. This vector contains text features such as length, punctuation score, and cognate score values. A set of manually prepared training data was assigned to train the probabilistic neural network and Gaussian mixture model. Another set of data was used for testing. Using the probabilistic neural network and Gaussian mixture model approaches, we could achieve error reduction of 27% and 50%, respectively, over the length based approach when applied on a set of parallel EnglishArabic documents. In addition, the results of (P-NNT) and (GMM) outperform the results of the combined model which exploits length, punctuation and cognates in a dynamic framework. The GMM approach outperforms Melamed and Moores approaches too. Moreover these new approaches are valid for any languages pair and are quite flexible since the feature vector may contain more, less or different features, such as a lexical matching feature and Hanzi characters in JapaneseChinese texts, than the ones used in the current research.
(Keyword)
sentence alignment / Englisb / Arabic parallel corpus / parallel corpora / probabilistic neural network / Gaussian mixture model / SPEAKER IDENTIFICATION / MODELS / WEB
Hua Xiang, Peilin Jiang, Shuang Xiao, Fuji Ren and Shingo Kuroiwa : A Model of Mental State Transition Network, IEEJ Transactions on Electronics, Information and Systems, Vol.127, No.3, 434-442, 2007.
(Summary)
Emotion is one of the most essential and basic attributes of human intelligence. Current AI (Artificial Intelligence) research is concentrating on physical components of emotion, rarely is it carried out from the view of psychology directly. Study on the model of artificial psychology is the first step in the development of human-computer interaction. As affective computing remains unpredictable, creating a reasonable mental model becomes the primary task for building a hybrid system. A pragmatic mental model is also the fundament of some key topics such as recognition and synthesis of emotions. In this paper a Mental State Transition Network Model is proposed to detect human emotions. By a series of psychological experiments, we present a new way to predict coming human's emotions depending on the various current emotional states under various stimuli. Besides, people in different genders and characters are taken into consideration in our investigation. According to the psychological experiments data derived from 200 questionnaires, a Mental State Transition Network Model for describing the transitions in distribution among the emotions and relationships between internal mental situations and external are concluded. Further more the coefficients of the mental transition network model were achieved. Comparing seven relative evaluating experiments, an average precision rate of 0.843 is achieved using a set of samples for the proposed model.
(Keyword)
artificial psychology / a mental state transition network / predict / psychological questionnaires
Satoru Tsuge, Shingo Kuroiwa, Masami Shishibori, Fuji Ren and Kenji Kita : Real-timeFrequency Characteristic Normalization for Distributed Speech Recognition, Transactions of Information Processing Society of Japan, Vol.48, No.2, 900-908, 2007.
Bracewell B. David, Fuji Ren and Shingo Kuroiwa : Mining News Sites to Create Special Domain News Collections, International Journal of Computational Intelligence, Vol.4, No.4, 56-63, 2007.
(Summary)
AbstractWe present a method to create special domain collections from news sites. The method only requires a single sample article as a seed. No prior corpus statistics are needed and the method is applicable to multiple languages. We examine various similarity measures and the creation of document collections for English and Japanese. The main contributions are as follows. First, the algorithm can build special domain collections from as little as one sample document. Second, unlike other algorithms it does not require a second general corpus to compute statistics. Third, in our testing the algorithm outperformed others in creating collections made up of highly relevant articles.
(Keyword)
Infromation Retrival / News / Special Domain Collections
305.
Shingo Kuroiwa, Yoji Mori, Masashi Takashina, Satoru Tsuge and Fuji Ren : Wind noise reduction method using the observed spectrum fine structure and esimated spectrum envelope, The Transactions of the Institute of Electronics, Information and Communication Engineers A, Vol.J90-A, No.1, 1-12, 2007.
Shuang Xiao, Hua Xiang, Fuji Ren and Shingo Kuroiwa : Extracting Chinese Idiomatic Expressions from Texts to Author Reading Support Systems for Learning Chinese as Second Language, The Journal of Information and Systems in Education, Vol.5, No.1, 17-28, 2006.
(Summary)
Though most current researches are focusing on Chinese reading support, there is still no system that can provide a convenient support aiming at CIE (Chinese Idiomatic Expression). In this paper, we mainly propose how to help CSL (Chinese as Second Language) learners to recognize CIE in Chinese texts by using NLP (Natural Language Processing) technique. We have created a CIE database with 2,305 idiomatic expressions of contemporary Chinese. The recognition processing of CIE consists of the CIE registration, the lexical analysis, the classification of sentences, the correction of POS tags and the extraction of CIE. In our experiment of CIE recognition, we have achieved 81.65% in Recall and 94.34% in precision. According to related researches of CIE recognition and comprehension, we created a RDBMS of CIE compre-hension support resources. For helping the learners to comprehend CIEs efficiently, we have designed a Support Program which connects the results of CIE recognition and the CIE comprehension support resources.
(Keyword)
Chinese Idiomatic Expression / phrase extraction
307.
Fattah Abdel Mohamed, Fuji Ren and Shingo Kuroiwa : Sentence Alignment using Feed Forward Neural Network, International Journal of Neural Systems, Vol.16, No.6, 423-434, 2006.
(Summary)
Parallel corpora have become an essential resource for work in multi lingual natural language processing. However, sentence aligned parallel corpora are more efficient than non-aligned parallel corpora for cross language information retrieval and machine translation applications. In this paper, we present a new approach to align sentences in bilingual parallel corpora based on feed forward neural network classifier. A feature parameter vector is extracted from the text pair under consideration. This vector contains text features such as length, punctuate score, and cognate score values. A set of manually prepared training data has been assigned to train the feed forward neural network. Another set of data was used for testing. Using this new approach, we could achieve an error reduction of 60% over length based approach when applied on English Arabic parallel documents. Moreover this new approach is valid for any language pair and it is quite flexible approach since the feature parameter vector may contain more/less or different features than that we used in our system such as lexical match feature.
Shingo Kuroiwa, Satoru Tsuge, Masahiko Kita and Fuji Ren : Evaluation of EMD-based Speaker Recognition using ISCSLP2006 Chinese Speaker Recognition Evaluation Corpus, Lecture Notes in Computer Science, Vol.4274, 539-548, 2006.
(Summary)
In this paper, we present the evaluation results of our proposed text-independent speaker recognition method based on the Earth Mover's Distance (EMD) using ISCSLP2006 Chinese speaker recognition evaluation corpus developed by the Chinese Corpus Consortium (CCC). The EMD based speaker recognition (EMD-SR) was originally designed to apply to a distributed speaker identification system, in which the feature vectors are compressed by vector quantization at a terminal and sent to a server that executes a pattern matching process. In this structure, we had to train speaker models using quantized data, so that we utilized a non-parametric speaker model and EMD. From the experimental results on a Japanese speech corpus, EMD-SR showed higher robustness to the quantized data than the conventional GMM technique. More- over, it has achieved higher accuracy than the GMM even if the data were not quantized. Hence, we have taken the challenge of ISCSLP2006 speaker recognition evaluation by using EMD-SR. Since the identification tasks were on an open-set basis, we introduce a new speaker verication module in this paper. Evaluation results showed that EMD-SR achieves 99.3% Identification Correctness Rate in a closed-channel speaker identication task.
Fuji Ren : Chinese Text Segmentation System Using Sensitive Word Concept, Journal of Asian Information-Science-Life, Vol.2, No.3, 209-222, 2006.
(Summary)
The identification of words in Indo-European languages is a trivial task. However, the problem named text segmentation has been, and is still a bottleneck for various Asian languages, such as Chinese. There have been two main groups of approaches to Chinese segmentation: dictionary-based approaches and statistical approaches. However, both approaches have difficulty to deal with some Chinese texts. To address the difficulties, we propose a hybrid approach using Sensitive Word Concept to Chinese text segmentation. Sensitive words are the compound words whose syntactic category is different from those of their components. According to the segmentation, a sensitive word may play different roles, leading to significantly different syntactic structures. In this paper, we explain the concept of sensitive words and their efficacy in text segmentation firstly, then describe the hybrid approach that combines the rule-based method and the probability-based method using the concept of sensitive words. Our experimental results showed that the presented approach is able to address the text segmentation problems effectively.
(Keyword)
Natural Language Processing / Chinese / Text Segmentation / Sensitive Words
310.
Fuji Ren : Machine-Aided English Writing Function in MMM Projest, Journal of Asian Information-Science-Life, Vol.2, No.3, 267-282, 2006.
(Summary)
Multi-lingual Multi-function Multi-media intelligent system (MMM) is a big project. It is a complex intelligent system with multiple functions that can deal with multiple languages and multiple media. We have designed a new general ontology based natural language processing system referred to as Multi-lingual Multi-function Multi-media Intelligent System. This paper describes a new Machine-Aided Writing Function in the MMM. The methodology proposed in this paper can deal with any foreign language, but we focus only on how to construct a Machine-Aided English Writing system for Japanese users. With the rapid development of the internet, writing English becomes daily work for computer users all over the world. However, for most non-native users, writing English is a big challenge. To build a machine-aided system that helps non-native users not only on spelling checking and grammar checking but also on producing accurate English expressions is a challenging task. The basic idea of the method proposed in this paper is based on Super-Function (SF) means and Corpora Intelligence Technique (CIT). SF is a new concept that we present to develop multi-lingual machine translation. A prototype Machine Aiding English Writing system has been constructed based on the proposed method. Some experiments using the prototype system have been carried out and the results show the proposed method is effective. This paper discusses the SF and CIT in detail and give some new advances in the MMM project.
Fuji Ren and Hongchi Shi : An Experimental parallel machine translation system, Journal of Asian Information-Science-Life, Vol.2, No.3, 223-242, 2006.
(Summary)
Many machine translation systems have been developed based on different machine translation approaches. The accuracy of machine translation can be further improved by integrating different machine translation systems into one system using parallel processing techniques. We have designed and implemented an experimental parallel machine translation (PMT) system. The system consists of four independent machine translation subsystems. Each subsystem is implemented using an existing machine translation technique and has its own characteristics. When translating a sentence, each subsystem translates it independently. If more than one subsystem translates the sentence successfully, the controller chooses the best translation according to a combining algorithm implemented using statistics collected from practical translations. If no subsystem succeeds in translating the sentence, the controller partitions the sentence into sentential parts, coordinates the subsystems to translates the sentential parts, and combines the partial translation results into a translation for the whole input sentence. In this paper, we describe our experimental PMT system, present some experimental results, and outline some future research on PMT.
Yixin Zhong and Fuji Ren : Mechanism Approach that Unifies AI, and AI with AE, WSEAS Transactions on Computers, Vol.5, No.10, 2204-2211, 2006.
(Summary)
Both Artificial Intelligence (AI) and Artificial Emotion (AE) have received much attention from academic as well as engineering circles. Structuralism, functionalism and behaviorism have been the three dominant approaches to the simulation of intelligence in history up to the present. All the approaches have made progress so far whereas they also, however, facing difficulties in their development. Further more, there have been no connections between the research in AI and that in AE. An attempt was thus made in the paper to propose another approach to the research in AI and AE. Different from the structuralism, functionalism and behaviorism, this approach is featured with the direct concern with the central mechanism of intelligence and emotion and is therefore termed mechanism approach. It is discovered as consequence that the transformations from information to knowledge and further to intelligence are the nucleus of the mechanism to AI and AE. A by-product that seems significant also achieved the same time that the three named approaches to AI can well be unified within the framework of mechanism approach. These discoveries, both AI and AE can successfully be simulated by mechanism approach and unifying the three approaches into harmonious one, may open up a new stage for the research in AI and AE as well as the integration of these two.
Kazuyuki Matsumoto, Fuji Ren and Shingo Kuroiwa : Emotion Estimation System based on Emotion Occurrence Sentence Pattern, Lecture Notes in Computer Science, Vol.4114, 902-911, 2006.
(Summary)
The approach of emotion estimation from the conventional text was for estimating superficial emotion expression mainly. However emotions may be included in humans utterance even if emotion expressions are not in it. In this paper, we proposed an emotion estimation algorithm for conversation sentence. We gave the rules of emotion occurrence to 1616 sentence patterns. In addition, we developed a dictionary which consisted of emotional words and emotional idioms. The proposed method can estimate emotions in a sentence by matching the sentence pattern of emotion occurrence and the rule. Furthermore, we can get two or more emotions included in the sentence by calculating emotion parameter. We constructed the experiment system based on the proposed method for evaluation. We analyzed weblog data including 253 sentences by the system, and conducted the experiment to evaluate emotion estimation accuracy. As a result, we obtained the estimation accuracy of about 60 %.
Dapeng Yin, Min Shao, Peilin Jiang, Fuji Ren and Shingo Kuroiwa : Treatment of Quantifiers in Chinese-Japanese Machine Translation, Lecture Notes in Computer Science, Vol.4114, 930-935, 2006.
(Summary)
Quantifiers and numerals often cause mistakes in Chinese-Japanese machine translation. In this paper, an approach is proposed based on the syntactic features after classification. Using the difference in type and position of quantifiers between Chinese and Japanese, quantifier translation rules were acquired. Evaluation was conducted using the acquired translation rules. Finally, the adaptability of the experimental data was verified and the methods achieved the accuracy of 90.75%, which showed that they were effective in processing quantifiers and numerals.
Jiajun Yan, BRACEWELL B. David, Fuji Ren and Shingo Kuroiwa : A Semantic Analyzer for Aiding Emotion Recognition in Chinese, Lecture Notes in Computer Science, Vol.4114, 893-901, 2006.
(Summary)
In this paper we present a semantic analyzer for aiding emotion recognition in Chinese. The analyzer uses a decision tree to assign semantic dependency relations between headwords and modifiers. It is able to achieve an accuracy of 83.5%. The semantic information is combined with rules for Chinese verbs containing emotion to describe the emotion of the people in the sentence. The rules give information on how to assign emotion to agents, receivers, etc. depending on the verb in the sentence.
In this paper, we present a new approach to align sentences in bilingual parallel corpora based on the use of the linguistic information of the text pair in Gaussian mixture model (GMM) classifier. A feature parameter vector is extracted from the text pair under consideration. This vector contains text features such as length, punctuation score, cognate score and a bilingual lexicon extracted from the parallel corpus under consideration. A set of manually prepared training data has been assigned to train the Gaussian mixture model. Another set of data was used for testing. Using the Gaussian mixture model approach, we could achieve error reduction of 160% over length based approach when applied on English-Arabic parallel documents. In addition, the results of (GMM) outperform the results of the combined model which exploits length, punctuation, cognate and bilingual lexicon in a dynamic framework.
Liying MI, Xin Luo, Fuji Ren and Shingo Kuroiwa : A Rule and Super Function-based Machine Translation System for Chinese-Japanese Causative Sentences, WSEAS Transactions on Computers, Vol.5, No.9, 2122-2129, 2006.
(Summary)
Machine translation (MT) can be generally classified as Rule-Based Machine Translation (RBMT) or Example-Based Machine Translation (EBMT) or Statistics-Based Machine Translation (SBMT), and each has numerous problems which remain unsolved. A rule-based machine translation system is an effective way to implement a machine translation system because of its extensibility and maintainability. However, it is generally disadvantageous in processing efficiency. In this paper, we present a hybrid approach based on fixed rules and Super Functions (SF) - Based method. SF is known for its lower costs, and is applicability to common users who do not have high demand on translation quality. Our purpose is to improve the processing efficiency and the quality of machine translation. In the present research, sufficient Chinese-Japanese causative sentence patterns have been employed as a language-database for experiment, which proves the suggested method can effectively improve translation quality within the range under discussion.
Yu Zhang, Fuji Ren and Shingo Kuroiwa : The Validity of Metaphor in Emotion Recognizing Model, WSEAS Transactions on Computers, Vol.5, No.9, 2049-2055, 2006.
(Summary)
Natural language has attracted importance to some researchers working on Emotion recognizing. We have made a Chinese emotion classification model and this time we add a new element to the emotion thesaurus of this model-metaphor. Through the studying of emotion, we find the metaphor is an important way of people's communication in natural language. In this paper, firstly, we introduce every part of our emotion classification model and are focus on the new element of metaphor in our emotion thesaurus. And then we do comparative experiments in order to test our model using metaphor before and after. At last the evaluation of our model is given and our model improved through using metaphor is proved to be a more effective model in recognizing emotion from text.
Dapeng Yin, Min Shao, Peilin Jiang, Fuji Ren and Shingo Kuroiwa : Rule-based Translation of Quantifiers for Chinese-Japanese Machine Translation, WSEAS Transactions on Computers, Vol.5, No.9, 2031-2036, 2006.
(Summary)
Quantifiers and numerals often cause mistakes in Chinese-Japanese machine translation. In this paper, an approach to quantifier translation is proposed based on the syntactic features after classification. First, morphological analysis is performed on sentences extracted from a Chinese-Japanese aligned corpus, which consists of quantifiers and numerals. Next, statistical information is obtained based on the meaning of the nouns with an accompanying quantifier. Using the difference in type and position of quantifiers between Chinese and Japanese, quantifier translation rules were acquired. Evaluation was conducted by using the acquired translation rules. Finally, the adaptability of the experimental data was verified and the methods achieved an accuracy of 90.75%, showing that they were effective in processing quantifiers and numerals translation.
Jiajun Yan, Bracewell B. David, Fuji Ren and Shingo Kuroiwa : An Integrated System for Semantic Analysis of Chinese, WSEAS Transactions on Computers, Vol.5, No.9, 1886-1891, 2006.
(Summary)
Semantic dependency analysis has extensive applications in Natural Language Processing (NLP). In this paper, we present an integrated system for semantic dependency analysis of Chinese. The system is composed of 3 main modules; Syntactic analysis, headword assignment, and semantic dependency assignment. For the semantic dependency module, many classifiers are tested. Primarily, we focus on assigning headwords and determining semantic relations between headwords and modifiers. For headword assignment, we handcrafted a set of rules and achieved a very high precision of 99.54%. For determining semantic relations, we experiment with four supervised machine learning algorithms with varying features for Headword-Modifier relations assignment, and achieved precision of 84%. The experimental results show that our approaches are an effective step in developing a semantic dependency analyzer for Chinese.
Bracewell B. David, Fuji Ren and Shingo Kuroiwa : Building Frames of Knowledge for Causal Agents in WordNet, WSEAS Transactions on Computers, Vol.5, No.9, 1880-1885, 2006.
(Summary)
WordNet has become a standard tool in the NLP researcher's toolkit. While giving a plethora of information it does lack certain information that would be of great benefit. This paper examines building frames of knowledge for a subset of causal agents in WordNet. This extra knowledge can help in Question & Answering, Machine Translation, etc. After an examination of the WordNet glosses different classes were created that allow for obtaining knowledge about actions, attributes, and domains.
Ya Lin, Tetsuya Tanioka, Kokichi Tanihira, Fuji Ren, Toshiko Tada, Katsuyo Howard and Haruo Kobayashi : An Interactive E-learning System for Practicing Team Care by Interdisciplinary Collaboration, Kawasaki Journal of Medical Welfare, Vol.12, No.1, 37-44, 2006.
(Summary)
"In recent years, remote education systems using telecommunication tools such as television and the internet have been developed and applied not only in open universities and preparatory schools, but also in areas where educational resources are scarce. However, these systems do not make immediate responses to questions from the learners in real time. In addition, an interactive education system connecting the computer system and the learners who need to practice team care by interdisciplinary collaboration has yet to be developed. That is the reason why we present ""interactive e-learning system for practicing team care by interdisciplinary collaboration"" and explore the possibility of introducing such a system into practice. As regards future considerations, it is important to develop effective learning content for this system. This content promotes and reinforces the essential attitude and skills for practicing interdisciplinary team care. Determination of the effectiveness of its application depends on whether the learners can practice interdisciplinary team care virtually and apply their learning to real situations."
Shunji Mitsuyoshi, Fuji Ren, Yasuto Tanaka and Shingo Kuroiwa : Non-Verbal Voice Emotion Analysis System, International Journal of Innovative Computing, Information and Control, Vol.2, No.4, 819-830, 2006.
(Summary)
A non-verbal voice analysis system that recognizes, separates and ranks concurrent emotions in real time has potential application in various fields, yet such a system that could delineate emotion based solely on the sound of human voice has not been successfully demonstrated before. Here, we propose a system that recognizes human emotion by means of analyzing the fundamental frequency of human voice taken from continuous natural speech. The system detects robust fundamental frequencies and intonations by parameterizing them into pitch, power, and deviation of power. Based on these parameters, data was classified via decision-tree logic into the emotional elements of anger, joy, sorrow, and calmness. Degree of excitement was also extracted. The system was evaluated by third parties by matching the system performance to human subjective classification for each element. Results indicate that overall matching rate was 70%, and the matching rate was 86% when compared to the subjects' assessment of their own voices. Our system performance exceeded the baseline with non-verbal information, which was equivalent to human subjective assessment.
Bracewell B. David, Fuji Ren and Shingo Kuroiwa : Dealing with Acronyms in Biomedical Texts, Engineering Letters, Vol.13, No.2, 216-224, 2006.
(Summary)
Recently, there has been a growth in the amount of machine readable information pertaining to the biomedical field. With this growth comes a desire to be able to extract information, answer questions, etc. based on the information in the documents. Many of these desired tasks require sophisticated language processing algorithms, such as part-of-speech tagging, parsing, and semantic interpretation. In order to use these algorithms the text must first be cleansed of acronyms, abbreviations, and misspellings. In this paper we look at identifying, expanding, and disambiguating acronyms in biomedical texts. We present an integrated system that combines previously used methods for dealing with acronyms and Natural Language Processing techniques in new way for a new domain. The result is an integrated system that achieves a high precision and recall. We break the task up into three modular steps: Identification, Expansion, and Disambiguation. During identification, each word is examined to determine if it is an acronym or not. For this,a hybrid approach that is composed of a Naive Bayesian classifier and a set of handcrafted rules is used. We are able to achieve results of 99.96% accuracy with a small training set. During the expansion step, a list of possible meanings for the words determined to be acronyms is created. We break the expansion up into two categories, local and global expansion. For local expansion we use windowing and longest common subsequence to generate the possible expansions. Global expansion requires an acronym database to retrieve the possible expansions.
(Keyword)
Acronyms / Text Cleansing / Bioinformatics
325.
Shuang Xiao, Hua Xiang, Fuji Ren and Shingo Kuroiwa : A CIE Extraction Syatem for CSL Learners, International Journal of Computer Science and Network Security, Vol.6, No.7A, 152-162, 2006.
(Summary)
Summary It is obliged to provide an effective reading support system for CSL (Chinese as Second Language) learners, for recognition and comprehension of CIE (Chinese Idiomatic Expression) in Chinese text are very difficult for them. Though most current research are focusing on Chinese reading support, there is still no perfect system that can provide a convenient support aiming at CIE. In this paper, we mainly propose how to help CSL learners recognize and comprehend CIE. We have created a CIE database with 2,305 idiomatic expressions of contemporary Chinese. At the same time we have analyzed the basic structures and using forms of these CIE. By the analysis we have presented an extraction approach which based on rules and characters of these CIE. In our extraction experiment of CCE, the recall achieved 81.65% and the precision achieved 94.34%.
326.
Haiqing Hu, Peilin Jiang, Fuji Ren and Shingo Kuroiwa : A New Question Answering System for Chinese Restricted Domain, IEICE Transactions on Information and Systems, Vol.E89-D, No.6, 1848-1859, 2006.
(Summary)
In this paper, we propose the construction of a web-based Question Answering (QA) system for restricted domain, which combines three resource information databases for the retrieval mechanism, including a Question&Answer database, a special domain documents database and the web resource retrieved by Google search engine. We describe a new retrieval technique of integrating a probabilistic technique based on OkapiBM25 and a semantic analysis which based on the ontology of HowNet knowledge base and a special domain HowNet created for the restricted domain. Furthermore, we provide a method of question expansion by computing word semantic similarity. The system is first developed for a middle-size domain of sightseeing information. The experiments proved the efficiency of our method for restricted domain and it is feasible to transfer to other domains expediently using the proposed method.
(Keyword)
question answering system / Web-based / semantic similarit / restricted domain / Chinese
Hua Xiang, Shuang Xiao, Fuji Ren and Shingo Kuroiwa : A Mind Model for an Affecitive Computer, International Journal of Computer Science and Network Security, Vol.6, No.6, 62-69, 2006.
(Summary)
A pragmatic emotional model is one of obligatory subjects for building a humanoid computer. This paper presents a conception of a Mental State Transition Network Model using psychological research for reference [1]. The ongoing work proposes a novel approach to detect human emotions. According to Pluchiks basic emotional classification, we defined nine basic emotional states and carried out a series of psychological investigations [2]. By the results of experiments we provide a new way to predict coming humans emotions depending on the various currents emotional states under various reinforcements. By means of statistic data derived from 227 questionnaires, the transitions in distribution among the emotions and relationships between internal mental situations and external stimuli are concluded. The high precision achieved in our evaluating experiments shows the model is helping in recognition and synthesis of human emotion.
328.
Zhong Zhang, Horoshi Toda, Hisanaga Fujiwara and Fuji Ren : Translation Invariant RI-spling Wavelet and Its Application on DE-noising, International Journal of Information Technology & Decision Making, Vol.5, No.2, 353-378, 2006.
(Summary)
Wavelet Shrinkage using DWT has been widely used in de-noising although DWT has a translation variance problem. In this study, we solve this problem by using the translation invariant DWT. For this purpose, we propose a new complex wavelet, the Real-Imaginary Spline Wavelet (RI-Spline wavelet). We also propose the Coherent Dual-Tree algorithm for the RI-Spline wavelet and extend it to the 2-Dimensional. Then we apply this translation invariant RI-Spline wavelet for translation invariant de-noising. Experimental results show that our method, when applied to ECG data, the medical image and the textile surface inspection can obtain better de-noising results than that of conventional Wavelet Shrinkage.
Fattah Abdel Mohamed, Fuji Ren and Shingo Kuroiwa : Effects of Phoneme Type and Frequency on Distributed Speaker Identification and Verification, IEICE Transactions on Information and Systems, Vol.E89-D, No.5, 1712-1719, 2006.
(Summary)
In the European Telecommunication Standards Institute (ETSI), Distributed Speech Recognition (DSR) front-end, the distortion added due to feature compression on the front end side increases the variance flooring effect, which in turn increases the identification error rate. The penalty incurred in reducing the bit rate is the degradation in speaker recognition performance. In this paper, we present a nontraditional solution for the previously mentioned problem. To reduce the bit rate, a speech signal is segmented at the client, and the most effective phonemes (determined according to their type and frequency) for speaker recognition are selected and sent to the server. Speaker recognition occurs at the server. Applying this approach to YOHO corpus, we achieved an identification error rate (ER) of 0.05% using an average segment of 20.4% for a testing utterance in a speaker identification task. We also achieved an equal error rate (EER) of 0.42% using an average segment of 15.1% for a testing utterance in a speaker verification task.
Zhao Xin, Fuji Ren and Shingo Kuroiwa : Translation of Japanese Noun Compounds at Super-Function Based MT System, IEEJ Transactions on Electronics, Information and Systems, Vol.126, No.5, 645-653, 2006.
(Summary)
The translation of noun compounds has become a major issue in Machine Translation (MT) due to their frequency of occurrence and high productivity. In our previous studies on Super-Function Based Machine Translation (SFBMT), we have found that noun compounds are very frequently used and difficult to be translated correctly, the overgeneration of noun compounds can be dangerous as it may introduce ambiguity in the translation. In this paper, we discuss the challenges in handling Japanese noun compounds in an SFBMT system, we present a shallow method for translating noun compounds by using a word level translation dictionary and target language monolingual corpus.
Yu Zhang, Zhuoming Li, Fuji Ren and Shingo Kuroiwa : A Preliminary Research of Chinese Emotion Classification Model, Research in Computing Science, Vol.19, 95-106, 2006.
(Summary)
There have been some studies about spoken natural language dialog, and most of them have successfully been developed within the specified domains. However, current human-computer interfaces only get the data to process their programs. Aiming at developing an affective dialog system, we have been exploring how to incorporate emotional aspects of dialog into existing dialog processing techniques. As a preliminary step toward this goal, we work on making a Chinese emotion classification model which is used to recognize the main affective attribute from a sentence or a text. Finally we have done experiments to evaluate our model.
332.
Shuang Xiao, Hua Xiang, Fuji Ren and Shingo Kuroiwa : The Recognition system of CCE for CSL Learners, Research in Computing Science, Vol.19, 49-61, 2006.
(Summary)
CSL (Chinese as Second Language) learners need relevant reading support because recognition of CCE (Chinese Conventional Expression) is a difficulty for them. In this paper, we mainly propose how to help CSL learners recognize CCE in Chinese text. We have created a CCE data-base with 2,305 conventional expressions of contemporary Chinese. At the same time we have analyzed the basic structures and application forms of these 2,305 conventional expressions. By the analysis we have presented an extraction approach which based on rules and characters of these CCE. In our extraction experiment of CCE, the recall achieved 81.65% and the precision achieved 94.34%.
333.
Fattah Abdel Mohamed, Fuji Ren and Shingo Kuroiwa : Stemming to Improve Translation Lexicon Creation form Bitexts, Information Processing & Management, Vol.42, No.4, 1003-1016, 2006.
(Summary)
Arabic is a morphologically rich language that presents significant challenges to many natural language processing applications because a word often conveys complex meanings decomposable into several morphemes (i.e. prefix, stem, suffix). By segmenting words into morphemes, we could improve the performance of English/Arabic translation pair's extraction from parallel texts. This paper describes two algorithms and their combination to automatically extract an English/Arabic bilingual dictionary from parallel texts that exist in the Internet archive after using an Arabic light stemmer as a preprocessing step. Before using the Arabic light stemmer, the total system precision and recall were 88.6% and 81.5% respectively, then the system precision an recall increased to 91.6% and 82.6% respectively after applying the Arabic light stemmer on the Arabic documents. The algorithms have certain variables which values can be changed to control the system precision and recall. Like most of the systems do, the accuracy of our system is directly proportional to the number of sentence pairs used. However our system is able to extract translation pairs from a very small parallel corpus. This new system can extract translations from only two sentences in one language and two sentences in the other language if the requirements of the system accomplished. Moreover, this system is able to extract word pairs that are translation of each others, synonyms and the explanation of the word in the other language as well. By controlling the system variables, we could achieve 100% precision for the output bilingual dictionary with a small recall.
Shingo Kuroiwa, Yoshiyuki Umeda, Satoru Tsuge and Fuji Ren : Nonparametric Speaker Recognition Method using Earth Mover's Distance, IEICE Transactions on Information and Systems, Vol.E89-D, No.3, 1074-1081, 2006.
(Summary)
In this paper, we propose a distributed speaker recognition method using a nonparametric speaker model and Earth Mover's Distance (EMD). In distributed speaker recognition, the quantized feature vectors are sent to a server. The Gaussian mixture model (GMM), the traditional method used for speaker recognition, is trained using the maximum likelihood approach. However, it is difficult to fit continuous density functions to quantized data. To overcome this problem, the proposed method represents each speaker model with a speaker-dependent VQ code histogram designed by registered feature vectors and directly calculates the distance between the histograms of speaker models and testing quantized feature vectors. To measure the distance between each speaker model and testing data, we use EMD which can calculate the distance between histograms with different bins. We conducted text-independent speaker identification experiments using the proposed method. Compared to results using the traditional GMM, the proposed method yielded relative error reductions of 32% for quantized data.
Shingo Kuroiwa, Satoru Tsuge, Koji Tanaka, Kazuma Hara and Fuji Ren : Acoustic Model Adaptation for Codec Speech based on Learning-by-Doing Concept, Advances in Natural Language Processing Research in Computing Science, Vol.18, 105-114, 2006.
(Summary)
Recently, personal digital assistants like cellular phones are shifting to IP terminals. The encoding-decoding process utilized for transmitting over IP networks deteriorates the quality of speech data. This deterioration causes degradation in speech recognition performance. Acoustic model adaptations can improve recognition performance. However,the conventional adaptation methods usually require a large amount of adaptation data. In this paper, we propse a novel acoustic model adaptation technique that generates speaker-independent HMM for the target environment based on the learning-by-doing concept. The proposedmethod uses HMM-based speech synthesis to generate adaptation datafrom the acoustic model of HMM-based speech recognizer, and consequently does not require any speech data for adaptation. By using thegenerated data after coding, the acoustic model is adapted to codecspeech. Experimental results on G.723.1 codec speech recognition showthat the proposed method improves speech recognition performance. Arelative word error rate reduction of approximately 12% was observed.
Satoru Tsuge, Masami Shishibori, Fuji Ren, Kenji Kita and Shingo Kuroiwa : Specific Speaker's Japanese Speech Corpus over Long and Short Time Periods, Advances in Natural Language Processing Research in Computing Science, Vol.18, 115-124, 2006.
(Summary)
We have been collecting the Japanese speech corpus for investigating the relationship between intra-speaker speech variability and speech recognition performance. In this paper, we introduce our speech corpus. Our corpus consists of six speakers speech data. Each speaker read specific utterance sets three times a day, once a week. Using a specific female speakers speech data in this corpus, we conduct speech recognition experiments for investigating the relationship between intra-speaker speech variability and speech recognition performance. Experimental results show that the variability ofrecognition performance over different days is larger than variability of recognition performance within a day.
337.
Fattah Abdel Mohamed, Fuji Ren, Shingo Kuroiwa, Satoru Tsuge and Ippei Fukuda : Phoneme Based Speaker Modeling to Improve Speaker Recognition System, Information, Vol.9, No.1, 135-147, 2006.
(Summary)
The most common approach to speaker recognition today is the use of global Gaussian Mixture Models (GMM) which ignores knowledge of the underlying phonetic content of the speech, so it does not take advantage of all available information. In this paper we investigate the phoneme effect on speaker recognition system. We found that some phonemes have strong effect on speaker identification. By segmenting the most effective phonemes for speaker recognition task from a speaker utterance, we could decrease the system complexity and the recognition time. Moreover, this technique is very useful to speed up the authentication process through wire/wireless communication systems. This paper is concerned with improving the performance of speaker recognition systems in two areas: decreasing the identification error rate and decreasing the utterance part required for identification task. We have applied several approaches on YOHO corpus; most of these approaches outperformed previously published results on the speaker identification task. Two of our approaches could achieve 0.0% error rate and one of these two approaches used only an average segment of 13% of the testing utterance for recognition.
(Keyword)
speaker recognition / phoneme effect on speaker recognition system / phonetic content
338.
Fattah Abdel Mohamed, Fuji Ren and Shingo Kuroiwa : Probabilistic Neural Network Based English-Arabic Sentence Alignment, Lecture Notes in Computer Science, Vol.3878, 97-100, 2006.
(Summary)
In this paper, we present a new approach to align sentences in bilingual parallel corpora based on a probabilistic neural network (P-NNT) classifier. A feature parameter vector is extracted from the text pair under consideration. This vector contains text features such as length, punctuation score, and cognate score values. A set of manually aligned training data was used to train the probabilistic neural network. Another set of data was used for testing. Using the probabilistic neural network approach, an error reduction of 27% was achieved over the length based approach when applied on English-Arabic parallel documents.
Haiqing Hu, Fuji Ren, Shingo Kuroiwa and Shuwu Zhang : A Question Answering System on Special Domain and the Implementation of Speech Interface, Lecture Notes in Computer Science, Vol.3878, 458-469, 2006.
(Summary)
In this paper, we propose a construction of Question Answering(QA) system, which synthesizes the answers retrieval from the frequent asked questions database and documents database, based on special domain about sightseeing information. A speech interface for the special domain was implemented along with the text interface, using an acoustic model HMM, a pronunciation lexicon, and a language model FSN on the basis of the feature of Chinese sentence patterns. We consider the synthetic model based on statistic VSM and shallow language analysis for sightseeing information. Experimental results showed high accuracy can be achieved for the special domain and the speech interface is available for frequently asked questions about sightseeing information.
(Keyword)
Question Answering System / Similarity Computing / Special Domain / speech recognition / Chinese
Mohamed Fattah, Fuji Ren and Shingo Kuroiwa : Speaker Recognition for Wire/Wireless Communication Systems, The International Arab Journal of Infromation Technology, Vol.3, No.1, 28-34, 2006.
(Summary)
Recently data communication spread to the mobile wireless world. The complexity of medium and large speech & speaker recognition systems are beyond the memory and computational resources of the small portable devices. Moreover the most common approach to speaker recognition today is the use of global Gaussian mixture models (GMM) which ignores knowledge of the underlying phonetic content of the speech, so it does not take advantage of all available information. In this paper we address the solution of these two problems by investigating the phoneme effect on speaker recognition system. We used YOHO database for speaker identification task. We found that some phonemes have strong effect on speaker identification. Segmenting the most effective phoneme for speaker recognition task from a speaker utterance and send this phoneme only through the wireless communication system will decrease the complexity of medium and speed up the authentication process though mobile communication system. We have applied different approaches on YOHO corpus, several of these approaches were able outperform previously published results on the speaker ID task. One of our approaches could achieve 0.7% error rate by using only an average segment of 4.45% of the testing utterance for recognition.
341.
Shingo Kuroiwa, Kuniyoshi Kobayashi, Satoru Tsuge and Fuji Ren : Packet Loss Concealment Algorithm using Speech Recognition and Synthesis, The Journal of the Acoustical Society of Japan, Vol.62, No.1, 3-11, 2006.
Fuji Ren and Kang Yen : Estimating the Minimum Entropy of Chinese and Japanese Languages, International Journal of Information Technology & Decision Making, Vol.4, No.4, 679-689, 2005.
(Summary)
The study of minimum entropy of a natural language has been an interesting research subject. For English a great progress has been made, but a few reports on other languages have been found in literature. Based on two hypotheses on conservation of information quantity, we proposed a method, which can be used to estimate the minimum entropy of characters in natural languages. With a large quantity of translation corpus, this method enables us to estimate the minimum entropy without calculating the probability. Besides, as the scale of translation corpus increases, the fluctuation of the ratio between character quantities in any two languages becomes negligible. In this paper, we apply this method to the study of two languages of a large character total, Japanese and Chinese.
(Keyword)
Minimum Entropy / Conservation of Information Quantity / Natural Language
Peilin Jiang, Hua Xiang, Fuji Ren and Shingo Kuroiwa : An Advanced Mental State Transtion Network and Psychological Experiments, Lecture Notes in Computer Science, Vol.3824, No.1, 1026-1035, 2005.
(Summary)
The study of human-computer interaction is now the mostpopular research domain overall computer science and psychology science. The most of essential issues recently focus on not only the information about the physical computing but also the affective computing.The emotion states of human being can dramatically affect their actions.It is important for a computer to understand what the people feel at thetime. In this paper, we propose a novel method to predict the future emotion state of person depending on the current emotion state and affectivefactors by an advanced mental state transition network. The psychological experiment with about 100 participants has been done to obtainthe structure and the coefficients of the model. The test experiment alsohas been done to certificate the prediction validity of this model.
Qiong Liu, Xin Lu, Fuji Ren and Shingo Kuroiwa : Automatic Stock Market Forecasting and Correlative Natural Language Generation, International Journal of Information, Vol.8, No.6, 881-890, 2005.
(Summary)
Time-series forecasting is an important research area in several domains. Recently, neural networks have been very successfully applied in time series to improve multivariate prediction ability. Several neural network model have already been developed for the market prediction. Some are applied to predicting the change of future interest rate and exchange rate; some are applied to recognizing certain price patterns that are characteristic of future price changes. This paper presents a neural network model for technical analysis of stock market, and its application to a buying and selling timing prediction system for stock index of Japan. This paper also describes a natural language generation system using Extensible Super-Function (ESF) to express prediction information of TOPIX in natural language for non-expert users. This system has evolved to be one of the most comprehensive grammars of English for prediction expressions.
345.
Ya Lin, Zhi Teng, Bracewell B. David, Fuji Ren and Shingo Kuroiwa : Development of a Multimedia Bidirectional Learning System Environment, International Journal of Information, Vol.8, No.6, 871-879, 2005.
(Summary)
During the last decade, multimedia learning system environments, using television and internet for instance, have been widely deployed not only in open universities, preparatory schools and technical exchange, but also in areas where the education system is underdeveloped. However, these learning systems are not capable of bidirectional learning. In this paper, we describe a bidirectional multimedia learning system environment. The bidirectional learning system environment is not the traditional e-Learning system, but is one that incorporates a surveillant system and a QA (Question&Answer) system. The surveillant system does not only track and record a learner's study status, but also prompts those who are not earnest at all times using a pattern recognition system. The QA system uses speech recognition and is used to identify and answer questions proposed by a learner. This system solves the problem of a lack of bidirectionality and control by a learner in the former e-Learning education systems.
(Keyword)
e-Learning system / Surveillant System / QA system / speech recognition system
346.
Qiong Liu, Xin Lu, Fuji Ren and Shingo Kuroiwa : XML based extended Super-Function schema in Knowledge Representation, Journal of Research on Computing Science, Vol.16, 3-12, 2005.
(Summary)
In recent years, the usual knowledge representation (KR) problem in artificial intelligence is how to automatically represent andtransform different kinds of knowledge using one kind of schema. Especially this problem focuses on representing formal knowledge in naturallanguage for human understanding. For this purpose, this paper proposesan extended super-function (ESF) schema to build a novel KR system.This system can translate the data of stock market or other fields intothe corresponding natural language expression automatically. Moreover,this system benefits from XML techniques which formalize and constructall information using the common Web rules to realize the ESF schema.
347.
Shingo Kuroiwa, Satoru Tsuge, Masami Shishibori, Fuji Ren and Kenji Kita : Simple PCAを用いたベクトル空間情報検索モデルの次元削減, IEEJ Transactions on Electronics, Information and Systems, Vol.125, No.11, 1773-1779, 2005.
(Summary)
In this paper, we propose to use the Simple Principal Component Analysis (SPCA) for dimensionality reduction of the vector space information retrieval model. The SPCA algorithm is a data-oriented fast method which does not require the computation of the variance-covariance matrix. In SPCA, principal components are estimated iteratively so we also propose a criteria to determine the convergence. The optimum number of iterations for each principal component can be determined using the criteria. Experimentally, we show that the SPCA-based method offers improvement over the conventional SVD-based method despite its small amount of computation. This advantage of SPCA can be attributed to its iterative procedure which is similar to clustering methods such as $k$-means clustering. On the other hand, the proposed method which orthogonalizes the basis vectors also achieved much higher accuracy than the conventional random projection method based on $k$-means clustering.
(Keyword)
情報検索 / ベクトル空間モデル / 次元削減 / Simple PCA / クラスタリング / Information retrieval / Vector Space Model / Dimensionality Reduction / Simple PCA / clustering
Bracewell B. David and Fuji Ren : A Memory and Search Hybrid Genetic Algorithm for non-Stationary Environments with Repetitive Natures, Research on Computing Science, Vol.14, No.1, 35-46, 2005.
(Summary)
We look at combining a search-based and memory-based approach creating a hybrid GA to solve problems with non-stationary environments. In particular, the memory search hybrid GA (MSHGA) we present is well suited to deal with non-stationary environments that are repetitive in nature, i.e. different problem landscapes are repeatedly seen. The MSHGA is capable of recalling candidate solutions for previously seen problem landscapes. This ability coupled with the search based technique of random immigrants causes the MSHGA to outperform the SGA and random immigrants GA in our experiments.
349.
Li Taihao and Fuji Ren : Super-Function based Ubiquitous Chinese Vocabulary Learning, INFORMATION, Vol.8, No.4, 547-556, 2005.
(Summary)
As intercourse and business between Japan and China increases, more Japanese go to China for traveling or for business. Overcoming the language obstacle is becoming an important project. This paper describes a ubiquitous Chinese language learning system using super-function (SF) in PDA. SF is a function that shows the correspondence between original and target language sentence patterns in machine translation. In his project, we focused on Chinese language learning to provide learners with the right information at the right time in the right way.
350.
Zhao Xin, Fuji Ren and Shingo Kuroiwa : Automatic Translation of Compound Nouns in the Japanese-Chinese Machine Translation System SFBMT, INFORMATION, Vol.8, No.3, 405-413, 2005.
(Summary)
The translation of compound nouns is a major issue in Machine Translation due to their frequency of occurrence and high productivity. Several aspects of compound nouns make them particularly difficult to be handled in a system performing automatic translation. In this paper, we discuss the challenges in automatic translating Japanese compound nouns into Chinese in the Super-Function based machine translation (SFBMT) system to address this issue. In our previous studies on SFBMT, we have found that there are many problems caused by compound nouns,the overgeneration of compound nouns can be dangerous as it may introduce ambiguity in the translation. To solve those problems, in this paper we present a shallow method for translating Japanese compound nouns into Chinese using a word level translation dictionary and target language monolingual corpus.
胡 海青, Fuji Ren and Shingo Kuroiwa : Chinese Automatic Question Answering System of Specific-Domain Based on Vector Space Model, IEEJ Transactions on Electronics, Information and Systems, Vol.125, No.5, 698-706, 2005.
(Summary)
In order to meet the demand to acquire necessary information efficiently from large electronic text, the Question and Answering (QA) technology to show a clear reply automatically to a question asked in the users natural language has widely attracted attention in recent years. Although the research of QA system in China is later than that in western countries and Japan, it has attracted more and more attention recently. In this paper, we propose a Question-Answering construction, which synthesizes the answer retrieval to the questions asked most frequently based on common knowledge, and the document retrieval concerning sightseeing information. In order to improve reply accuracy, one must consider the synthetic model based on statistic VSM and the shallow semantic analysis, and the domain is limited to sightseeing information. A Chinese QA system about sightseeing based on the proposed method has been built. The result is obtained by evaluation experiments, where high accuracy can be achieved when the results of retrieval were regarded as correct, if the correct answer appeared among those of the top three resemblance degree. The experiments proved the efficiency of our method and it is feasible to develop Question-Answering technology based on this method.
(Keyword)
Question Answering System / Information Retrieval / Vector Space Model / Chinese / Vector Space Model / Chinese
近年，クリニカルパスはケアを改善するために用いられるようになった．この研究の目的は統合失調症用のためのCPを開発することである．その主要な焦点は，①ノーマライゼーションの理念に基づいたチームケアを実践すること，②CPと連動した患者アウトカム管理用の重要な品質管理項目を明確にすることである．アウトカム管理を病院で行うことによって，患者·家族のQuality of Lifeの向上，良好な費用対効果，医療者の満足度の向上，入院期間の短縮等の効果が得られる．筆者らは，(Interdisciplinary Collaborative Team Care Model: ICTCM)を2000年に導入し，1999年の平均在院日数205．7日(1994年の平均在院日数675．5日)を2004年には155．2日に短縮した病院において，学際的なケア提供者と事務責任者を調査対象として，聞き取り調査を行い，統合失調症用のCPとアウトカム管理項目を作成した．
(Keyword)
schizophrenia / Normalization / Outcome management / Clinical path / Team care
Fuji Ren : Automatic abstracting important sentences, International Journal of Information Technology & Decision Making, Vol.4, No.1, 141-152, 2005.
(Summary)
Being increasingly popular, the Internet greatly changes our lives. We can conveniently receive and send information via the Internet. With the information explosion on the Web, it is becoming crucial to develop means to automatically extract important sentences from the Web articles. In this paper, we propose a method which uses both statistical and structural information for sentence extraction. In addition, following the analysis of human's extractions, several heuristic rules are added to filter out non-important sentences and to prevent similar sentences from being extracted. Our experimental results proved the effectiveness of these means. In particular, once the heuristic rules being added, a significant improvement has been observed.
(Keyword)
Automatic Extraction of Important Sentence / statistical information / structural feature / web
Wang Xiaojie and Fuji Ren : Chinese-Japanese Clause Alignment, Lecture Notes in Computer Science, Vol.3406, No.1, 400-412, 2005.
(Summary)
Bi-text alignment is useful to many Natural Language Processing tasks such as machine translation, bilingual lexicography and word sense disambiguation. This paper presents a Chinese-Japanese alignment at the level of clause. After describing some characteristics in Chinese-Japanese bilingual texts, we first investigate some statistical properties of Chinese-Japanese bilingual corpus, including the correlation test of text lengths between two languages and the distribution test of length ratio data. We then pay more attention to n-m(n>1 or m>1) alignment modes which are prone to mismatch. We propose a similarity measure based on Hanzi characters information for these kinds of alignment mode. By using dynamic programming, we combine statistical information and Hanzi character information to find the overall least cost in aligning. Experiments show our algorithm can achieve good alignment accuracy.
(Keyword)
Parallel corpus / alignment / Natural Language Processing
Hu Haiqing, Fuji Ren and Shingo Kuroiwa : Chinese Automatic Question Answering System for Sightseeing Tourists, International Journal of Information, Vol.8, No.1, 177-186, 2005.
(Summary)
In this paper, we propose a Question-Answering (QA) construction, which synthesizes the answers retrieved to the questions most frequently asked, based on common knowledge and text knowledge about sightseeing information. In order to improve replies more accurately, one must consider the synthetic model based on statistic VSM and the shallow semantic analysis, so that the domain is only limited to sightseeing information. Finally, the result is obtained by evaluation experiments, where high accuracy can be achieved when the results of retrieval are seen as correct, if the correct answer has appeared up to the third higher rank which sorts the value of the resembling degree. The experiments proved the efficiency in our method and it is feasible to use this method to develop Question-Answering technology.
(Keyword)
Question Answering System / Sightseeing Tpurists / text knowledge / vector space model
356.
Fattah Abdel Mohamed, Fuji Ren and Shingo Kuroiwa : Adaptive Threshold Parameters for Bilingual Dictionary Extraction from the Internet Archive, International Journal of Information, Vol.8, No.1, 165-176, 2005.
(Summary)
Parallel corpus is a very important tool to construct a good machine translation system or make any natural language processing research for cross language information retrieval. Internet archive is a good source of parallel documents of different languages. In order to construct a good parallel corpus from the Internet archive Bilingual dictionary is a must. This paper describes two algorithms to automatically extract an English/ Arabic bilingual dictionary from parallel texts that exist in the Internet archive. Each algorithm has certain parameters which values could be changed to increase the system precision and the number of extracted translation pairs. The system should preferably be useful for many different language pairs. Like most of t! he systems done, the accuracy of our system is directly proportional to the number of sentence pairs used. But our system is able to extract translation pairs from a very small parallel corpus. This new system can extract translations from only two sentences in one language and two sentences in the other language if the requirements of the system accomplished. Moreover, this system is able to extract word pairs that are translation of each other and the explanation of the word in the other language as well. We applied the system on English / Arabic parallel documents and the results were good. By controlling the system parameters, we could achieve very high precision for the output bilingual dictionary.
(Keyword)
Parallel corpus / machine translation system / Bilingual dictionary
357.
Fuji Ren : 人間感情の認知と機械感情の創生ができる感情インターフェース, International Journal of Information, Vol.8, No.1, 7-20, 2005.
Zhao Xin, Fuji Ren and Vob Stefan : A Super-Function Based Japanese-Chinese Machine Translation System for Business Users, Lecture Notes in Computer Science, Vol.3265, No.1, 272-281, 2004.
(Summary)
In this paper, a Japanese-Chinese machine translation system using the so-called Super-Function (SF) approach is presented. A SF is a functional relation mapping sentences from one language to another. The core of the system uses the SF approach to translate without going through syntactic and semantic analysis as many MT systems usually do. Our work focuses on business users for whom MT often is a great help if they need an immediate idea of the content of texts like e-mail messages, reports, web pages, or business letters. In this paper, we aim at performing MT between Japanese and Chinese to translate business letters by the SF based technique.
(Keyword)
Super-Function / business letters / machine translation system
Nasiff-Hada G., Yen K., Caballero A. and Fuji Ren : Collective Behavior of Distributed Systems in Manufacturing Environments, Engineering Mechanics, Vol.11, No.4, 1-10, 2004.
(Summary)
We study the Hogg-Huberman model used for resource allocation problem and extend the results of 2 agents and 2 resources to L agents and L resources. It has been shown that when processes can choose among many possible strategies, which collaborating in the solution of tasks, the dynamics can lead to oscillations and chaos. In this paper a first-order exponential filter has been proposed to improve the response. Another solution based on the average of the payoff functions for every resource shared by agents can also achieve a stable response and hold a good convergence time.
(Keyword)
resource allocation problem / Hogg-Huberman model / solution of tasks
360.
Fuji Ren : An Algorithm for Determining DingYu Structural Particle DE, Computational Linguistics and Intelligent Text Processing, Vol.2945, No.1, 338-349, 2004.
(Summary)
Nowadays a practical Japanese-Chinese machine translation system based on translation rules has been implemented. However, the current system lacks the ability for resolving the problem mentioned above. To resolve this problem, this paper presents an algorithm for determining DingYu structural particle using the grammar knowledge and statistical information. We first collect a large number of grammar items from Chinese grammar books, and obtain some elementary judgment rules by classifying and inducing the collected grammar items. Then we put these judgment rules into use in actual Chinese language, and modify the rules by checking their results instantly. Lastly we check and modify the rules by using the statistical information from a actual corpus. An experiment system based on the proposed algorithm has been constructed and an experiment is carried out. The result shows the effectiveness of the presented method.
Fuji Ren and Shunji Mitsuyoshi : To Understand and Create the Emotion and Sensitivity, International Journal of Information, Vol.6, No.5, 547-556, 2003.