Home      About Journal      Editorial Board      Instructions      Ethics Statement      Contact Us                   中文

Judgement and Disposal of Academic Misconduct Article Copyright Transfer Agreement Proof of Confidentiality Requirements for Electronic Version More.... Chinese Association of Automation National ResearchCenter for Intelligent Computing System Institute of Intelligent Machines,Chinese Academy of Sciences More....

2013 Vol.26 Issue.11, Published 2013-11-30 Papers and Reports    Researches and Applications        Papers and Reports 993 Covariance Modeling and Bayesian Classification of Halftone Image on Riemannian Manifold WEN Zhi-Qiang,HU Yong-Xiang,ZHU Wen-Qiu A covariance modeling method and a Bayesian method on Riemannian manifold are presented for classification of halftone image. According to the Fourier spectrum characteristic of halftone image,a feature extraction based on template matrices is presented to form a covariance matrix by combining with the spectrum of halftone image. An algorithm for covariance matrix extraction of halftone image is proposed by introducing a decision rule of effective image and partitioning technology. A Bayesian rule based on neighbor characteristic of tested samples and kernel density estimation is presented on Riemannian manifolds of symmetric positive definite matrices. In experiments,the problem of selection on threshold parameter is studied by statistical methods,the comparisons of the proposed method with 5 similar methods are conducted,and the influences of two parameters on classification performance and time cost of feature modeling are discussed. The experimental results show that the classification error of the proposed method is below 4% and computation time of modeling is under 100ms if parameters Q=32 or 64 and L=10~15. Furthermore,the proposed method is superior to other 5 methods. 2013 Vol. 26 (11): 993-1003 [Abstract] ( 476 ) [HTML 1KB] [ PDF 1730KB] ( 910 ) 1004 Cross-Domain Sentiment Analysis Based on Weighted SimRank WEI Xian-Hui,ZHANG Shao-Wu,YANG Liang,LIN Hong-Fei Cross-domain sentiment classification has attracted more attention in natural language processing field currently. It aims to predict the text polarity of target domain with the help of labeled texts in source domain. Usually,traditional supervised classification approaches can not perform well due to the difference of data distribution between domains. In this paper,a weighted SimRank algorithm is proposed to address this problem. The weighted SimRank algorithm is applied to construct a Latent Feature Space (LFS) with feature similarity. Then each sample is reweighted by the mapping function learned from the LFS. After reducing the mismatch of data distribution between domains,the algorithm performs well on cross-domain sentiment classification. The experiment verifies the effectiveness of the proposed algorithm. 2013 Vol. 26 (11): 1004-1009 [Abstract] ( 504 ) [HTML 1KB] [ PDF 405KB] ( 992 ) 1010 Research on Interval Reduction Model in Variable Precision Rough Set SONG Xiao-Wei,WANG Jia-Yang Interval reduction models based on classification quality and positive region lead to different kinds of reduction anomalies in variable precision rough set model (VPRS-Model). The reason is that the size of condition classification changes with the reduction of condition attributes,besides,classification quality,positive region and lower approximation distribution do not change equivalently any more. A reduction model based on lower approximation distribution is defined to avoid all kinds of reduction anomalies and an interval reduction method is presented based on ordered discernibility matrix. At last,the application of 3 kinds of interval reduction model on Wine Dataset illustrates the relationship of different reduction models. 2013 Vol. 26 (11): 1010-1018 [Abstract] ( 318 ) [HTML 1KB] [ PDF 407KB] ( 456 ) 1019 Trainable Emotional Speech Synthesis Based on PAD CHEN Yan-Xiang,LONG Run-Tian Emotional speech synthesis is the emphasis and hotspot in affective computing and speech signal processing. In speech synthesis,accurate speech emotion analysis is a prerequisite for high-quality synthesis of emotional speech. In this paper,PAD emotional model is used to build a 3D emotional space for sentiment analysis and clustering of emotional corpus of speech in order to get emotional PAD parameters model. The emotional speech is synthesized by HMM speech synthesis system,and the parameters of synthesized speech emotion are modified by PAD model. Therefore,the quality of emotional speech synthesis is improved. The experimental results show that the proposed method improves the naturalness of synthesized speech and the clarity of emotion and also achieves good performance among different male speakers. 2013 Vol. 26 (11): 1019-1025 [Abstract] ( 490 ) [HTML 1KB] [ PDF 468KB] ( 819 ) 1026 Online Local Adaptive Fuzzy C-Means Clustering Algorithm WU Xiao-Yan,CHEN Song-Can An online local adaptive fuzzy C-means (OLAFCM) algorithm for high dimensional data is proposed based on fuzzy C-means (FCM) and local adaptive clustering (LAC). Through assigning corresponding weights to its attributes,OLAFCM can make each cluster distribute in a subspace spanned by the combination of different attributes. Thus,the proposed algorithm not only avoids the risk of loss of information encountered in global dimensionality reduction techniques,but also is suitable for clustering data streams. Compared to state-of-the-art partition-based online clustering algorithms using global dimensionality reduction methods,the proposed algorithm has better performance on artificial and real datasets. 2013 Vol. 26 (11): 1026-1032 [Abstract] ( 320 ) [HTML 1KB] [ PDF 591KB] ( 534 ) 1033 A Variational Level Set Method for Image Segmentation Based on Improved Signed Distance Function CUI Yu-Ling In order to maintain the stability of traditional level set methods,the re-initialization method or a signed distance function is often used. However,those two methods are either time-consuming or instable. Thus,a signed distance function based level set method is proposed for overcoming those disadvantages. Firstly,the existing Double-Well constraint term is improved,which avoids re-initialization,increases the computational efficiency and makes the evolution more stable. Secondly,the active contour model based on the global grey information and local grey information is used to construct the energy function,thus it inherits the advantages of global and local models and drives the level set function accurately to real objective boundaries. Besides,the weigh of combination can be adjusted dynamically. At last,Gaussian convolution is presented to accelerate the speed of evolution and smooth the level set function. The experiments on both synthesis images and real images show that the proposed method has high computational efficiency and accuracy,and it is robust to noise and initial contour. 2013 Vol. 26 (11): 1033-1040 [Abstract] ( 431 ) [HTML 1KB] [ PDF 780KB] ( 1201 ) Researches and Applications 1041 Cooperative Co-Evolutionary Cuckoo Search Algorithm for Continuous Function Optimization Problems HU Xin-Xin,YIN Yi-Long To improve the performance of cuckoo search algorithm for continuous function optimization problems,a cooperative co-evolutionary cuckoo search algorithm is proposed. Through the framework of cooperative co-evolutionary,the improved algorithm divides the solution vectors of population into several sub-vectors and constructs the corresponding sub-swarms. The solution vectors of each sub-population are updated by the standard cuckoo search algorithm. Each sub-population provides the vectors of the best solution,which are combined with solution vectors of other sub-populations,and the combined solution vectors are evaluated. The simulation experiments on 10 benchmark functions show that the proposed algorithm efficiently improves the performances on contnuous function optimization problems and it is a competitive optimization algorithm for the problems compared with other algorithms. 2013 Vol. 26 (11): 1041-1049 [Abstract] ( 488 ) [HTML 1KB] [ PDF 555KB] ( 802 ) 1050 Image Retrieval with Spatial Context Weighting Based Vocabulary Tree ZHU Dao-Guang,GUO Zhi-Gang,ZHAO Yong-Wei Vocabulary tree based Bag-of-Words (BoW) representation becomes popular for image retrieval recently. Aiming at the absence of spatial context information in conventional vocabulary tree approaches,an image retrieval approach using spatial context weighting based vocabulary tree is proposed. Within the framework of vocabulary tree,this approach firstly describes the spatial context information of SIFT features. Then,the matching scores between SIFT features are weighted based on spatial context similarity,and similarities between images are achieved. Finally,image retrieval results are obtained according to the ranking of similarities. The experimental results indicate that the retrieval performance is improved and the proposed approach applies to large scale databases. 2013 Vol. 26 (11): 1050-1056 [Abstract] ( 385 ) [HTML 1KB] [ PDF 405KB] ( 1689 ) 1057 Dynamic Double Subgroups Cooperative Fruit Fly Optimization Algorithm HAN Jun-Ying,LIU Cheng-Zhong,WANG Lian-Guo In order to overcome the demerits of basic Fruit Fly Optimization Algorithm(FOA),such as low convergence precision and easily relapsing into local optimum,a dynamic double subgroup cooperative Fruit Fly Optimization Algorithm (DDSCFOA) is presented. Firstly,the whole group is dynamically divided into advanced subgroup and backward subgroup according to its own evolutionary level. Secondly,a finely local searching is made for advanced subgroup in the neighborhood of local optimum with Chaos algorithm,and a global search with FOA is made for backward subgroup,so that the whole group keeps in good balance between the global searching ability and local searching ability. Finally,two subgroups exchange information by updating the overall optimum and recombining the subgroups. DDSCFOA can jump out of local optimum and avoid falling into local optimum. The experimental results show that the strategy of dynamic double subgroup cooperative evolution is effective and feasible,DDSCFOA is much better than basic FOA in convergence velocity and convergence precision. 2013 Vol. 26 (11): 1057-1067 [Abstract] ( 446 ) [HTML 1KB] [ PDF 800KB] ( 776 ) 1068 Feature Selection for Cross-Domain Sentiment Classification ZHANG Yu-Hong,ZHOU Quan,HU Xue-Gang The data is usually unlabeled in application,which makes the adaptation of cross-domain effective. However,the sentiment classification is domain-dependent. The feature space of source domain,gotten by feature selection,can not represent the common character of both domains and is not suitable for the classification of target domain. Therefore,an approach of feature selection for cross-domain sentiment classification,Log-Likelihood Ratio-Term Frequency (LLRTF) is proposed. The log likelihood ratios (LLR) of features are computed in source domain,by which the discriminative feature space is gotten. Then,the statistic information term frequency of both domains is added to the LLR,and the features which are more important in target domain are selected. The feature space construction based on the LLRTF reduces the difference between source domain and target domain. The experimental result shows that the LLRTF is superior to the baselines. 2013 Vol. 26 (11): 1068-1072 [Abstract] ( 340 ) [HTML 1KB] [ PDF 429KB] ( 609 ) 1073 Face Image Analysis Based on Multiple Separated Component Sparse Coding LIU Wei-Feng,LIU Hong-Li,WANG Yan-Jiang Considering the different contributions of different facial components to face analysis,e.g. eyes,mouth etc.,a face analysis based on multi-component sparse coding is proposed. Firstly,some facial components which play important role to face analysis are selected. Then,the dictionaries of multiple components are learnt by using multi-view sparse coding algorithm,and the sparse codes of each face image are computed based on the dictionary. The final decision is made through pooling the sparse codes into support vector machines and least squares classifiers. Face analysis experiments include face recognition,facial expression recognition,face recognition with occlusion,and facial expression recognition with occlusion. The experimental results show that the proposed method based on multi-component sparse coding learns optimal weights of different facial components and outperforms single facial component method and simple multi-component fusion method. 2013 Vol. 26 (11): 1073-1078 [Abstract] ( 390 ) [HTML 1KB] [ PDF 709KB] ( 615 ) 1079 An Improved Wang-Mendel Method Based on Cooperation Degree of Sample and Self-Organizing Mapping GOU Jin,CHEN Wen-Yu Wang-Mendel algorithm is commonly used as a classic method to generate fuzzy rule base. But rules with low confidence are usually extracted when noise appears in the sample data set,while its efficiency also often drops fast when the scale of sample data increases. To solve those problems,two methods,cooperation relationship and self-organizing mapping (SOM) neural network,are introduced. Cooperation relationship among sample data improves the accuracy of rules and approximation ability to the original model. On the other hand,SOM can well preprocess sample data for denoising and reduce its scale through a self-adaptive learning procedure of weights network. Then an improved Wang-Mendel algorithm is proposed based on cooperation relationship degree of sample data and SOM. The experimental results,including trigonometric function approximation and artificial driving simulation of a train operation control system,show its completeness,robustness and operating efficiency. 2013 Vol. 26 (11): 1079-1085 [Abstract] ( 460 ) [HTML 1KB] [ PDF 1433KB] ( 595 )

Supervised by China Association for Science and Technology Sponsored by Chinese Association of Automation NationalResearchCenter for Intelligent Computing System Institute of Intelligent Machines, Chinese Academy of Sciences Published by Science Press

Copyright © 2010 Editorial Office of Pattern Recognition and Artificial Intelligence Address: No.350 Shushanhu Road, Hefei, Anhui Province, P.R. China Tel: 0551-65591176 Fax:0551-65591176 Email: bjb@iim.ac.cn Supported by Beijing Magtech 　Email:support@magtech.com.cn