模式识别与人工智能
Home      About Journal      Editorial Board      Instructions      Ethics Statement      Contact Us                   中文
Pattern Recognition and Artificial Intelligence
22 Judgement and Disposal of Academic Misconduct Article
22 Copyright Transfer Agreement
22 Proof of Confidentiality
22 Requirements for Electronic Version
More....
22 Chinese Association of Automation
22 National ResearchCenter for Intelligent Computing System
22 Institute of Intelligent Machines,Chinese Academy of Sciences
More....
 
 
2014 Vol.27 Issue.3, Published 2014-03-30
Papers and Reports    Researches and Applications    Surveys and Reviews   
   
Papers and Reports
193 Information Entropy Ensemble Classification Algorithm for Incomplete Data
ZHAO Shu,Lv Jing,ZHANG Yan-Ping,ZHANG Yi-Wen
Ensemble method is a simple and effective method to deal with incomplete data for classification. However,the weight of each sub-classifier in ensemble classification algorithm for incomplete data is mainly determined by the size and dimension of corresponding sub-dataset at present. The contributions of the missing attributes are different,and information entropy is introduced to measure these differences,thus,a novel algorithm for incomplete data named Entropy Ensemble Classification Algorithm (EECA) is proposed in this paper. The ensemble classifier with BP neural network being base classifier is applied on UCI dataset. The experimental results show that EECA determining the weight for sub-classifier by information entropy is better than the algorithm by using simple weight.
2014 Vol. 27 (3): 193-198 [Abstract] ( 611 ) [HTML 1KB] [ PDF 0KB] ( 151 )
199 Fuzzy Discriminant Analysis Based on Sparse Similarity Measurement
SONG Xiao Ning,XU Yong
The mathematic essence of sparse representation is signal decomposition under the constraint of sparsity regularization. A fuzzy discriminant analysis based on sparse similarity measurement is proposed in this paper. Each high dimensional image sample is firstly partitioned into several local blocks with equal size by the proposed algorithm,and these local blocks are combined to represent the samples as a Ridgelet sequence. Then,a new sparse learning algorithm is presented for coefficient decomposition and the corresponding sparse similarity measurement,and the fuzzy discriminant analysis criterion is subsequently developed by embedding the sparse similarity. The proposed algorithm successfully utilizes the novel sparse supervised learning algorithm as a feature extraction tool. Meanwhile ,it overcomes the shortcomings of traditional discriminant analysis method derived from the lack of structure knowledge between samples,especially in the case of high dimensional nonlinear small sample sizes. The experimental results on the ORL and FERET face images show the effectiveness of the proposed method.
2014 Vol. 27 (3): 199-205 [Abstract] ( 427 ) [HTML 1KB] [ PDF 0KB] ( 91 )
206 Gene Read Mapping Algorithms Based on MapReduce
TU Jin-Jin,YANG Ming,GUO Li-Na
Massive data generated by the rapid development of RNA-seq sequencing technology make serious challenges to the original read mapping algorithm in the efficiency. A spaced seed indexing algorithm without considering splice site based on MapReduce(PSeqMap),a spaced seed indexingalgorithm considering splice site(PJuncSeqMap),and a load-balancing solution are proposed. The MapReduce framework is employed to parallelize spaced seed indexing algorithms. The experimental results on the Arabidopsis gene datasets show that the proposed algorithms take full advantage of storage and computing power of the clusters and process massive genetic data efficiently.
2014 Vol. 27 (3): 206-212 [Abstract] ( 453 ) [HTML 1KB] [ PDF 0KB] ( 162 )
Surveys and Reviews
213 A Survey on Head Pose Estimation
TANG Yun-Qi,SUN Zhe-Nan,TAN Tie-Niu
Lots of head pose estimation methods have been presented,while there is not any survey that systematically summarizes the existing methods. To address this problem,detailed description of head pose estimation is firstly presented to define what head pose estimation is and how to describe it. Then,the existing head pose estimation methods are categorized according to the data source used by the methods,the automation of the methods and the principle of the methods. Furthermore,the principle and data source are taken as the main line and auxiliary line respectively to detailedly review and analyze the existing head pose estimation methods.
2014 Vol. 27 (3): 213-225 [Abstract] ( 817 ) [HTML 1KB] [ PDF 0KB] ( 94 )
Researches and Applications
226 Texts Similarity Algorithm Based on Subtrees Matching
ZHANG Pei-Yun,CHEN Chuan-Ming,HUANG Bo
To reduce the dimensionality of text vectors and improve the performance of semantic similarity measurement,an algorithm for texts similarity computation is proposed,which combines the advantages of the statistical methods and semantic dictionary. The texts are utilized to generate metadata feature vectors,so that it reduces the dimensionality of text vectors space. The algorithm for computing texts similarity is designed based on subtrees matching and the speed of computing texts similarity is improved. The accuracy of texts semantic similarity measurement is improved by utilizing the semantic matching of metadata feature vectors and subtrees. The synonyms widely existing in metadata are processed by the proposed method,and the semantic coverage ability for similarity computation of texts is also enhanced. The experimental results show that the proposed method is feasible and effective.
2014 Vol. 27 (3): 226-234 [Abstract] ( 501 ) [HTML 1KB] [ PDF 0KB] ( 123 )
235 A Recognition Method for Distorted and Merged Text-Based CAPTCHA
YIN Long,YIN Dong,ZHANG Rong,WANG De-Jian
The study of CAPTCHA recognition can discover CAPTCHA security vulnerabilities in time to make it more secure. Distorted and merged CAPTCHA can resist character segmentation,which is the difficult in CAPTCHA recognition. An approach based on DENSE SIFT and RANSAC algorithm is presented for recognition of distorted and merged CAPTCHA. Firstly,matching set is obtained through the matching of DENSE SIFT. Then,matching information is got by using RANSAC algorithm. Finally,recognition results are acquired by means of queue-analysis algorithm. The experimental results show that the proposed method has good performance on CAPTCHAs in different levels of difficulty.
2014 Vol. 27 (3): 235-241 [Abstract] ( 819 ) [HTML 1KB] [ PDF 0KB] ( 76 )
242 Low-Dimensional Histogram of Oriented Gradients with Non-Overlapping Scheme
HUO Ya-Song,ZHANG Kun
As a derivation version of scale-invariant feature transform (SIFT),histogram of oriented gradients (HOG) is widely used in human detection,gesture recognition,face recognition,scene classification,etc. However,the high dimension of the HOG feature vector leads to the curse of the dimensionality and high computation complexity. In this paper,it is found that the high dimension of HOG feature vector results from computing histograms of overlapping blocks. Though overlapping block is useful for enhancing the robustness,it leads to redundant information. To reduce the redundant information and the number of features as well,a non-overlapping version of HOG is proposed. The dimensions of the proposed method are 1/3 of those of traditional ones. The experimental results on palm and human detection demonstrate the efficiency and effectiveness of the proposed method.
2014 Vol. 27 (3): 242-247 [Abstract] ( 436 ) [HTML 1KB] [ PDF 0KB] ( 91 )
248 An Improved Markov Random Field Classification Approach for Hyperspectral Data Based on Efficient Belief Propagation
CAO Yang ,ZHAO Hui-Jie ,HUANG Si-Niu ,LI Na,ZHANG Pei
Aiming at the problems of imprecise class conditional probability (CCP) estimation and heavy computational cost for the global energy minimum in Markov random field (MRF) based classification algorithm,an improved MRF approach based on efficient belief propagation (EBP) is developed for land-cover classification of hyperspectral data. The estimation accuracy of the CCP is improved by the probabilistic support vector machine (PSVM) algorithm using spectral information of pixels,then the spatial correlation information is introduced by the MRF classification algorithm,thus the spectral information and spatial information is combined effectively. Moreover,an EBP optimization algorithm is developed,by which the computational cost is reduced and the classification accuracy is improved. The experimental results show that the proposed approach is effective. The average classification accuracy is up to 95.78%,Kappa coefficient is 93.34%,and the computational time of EBP is about 25% of that by belief propagation algorithm. Therefore,the proposed approach is valuable in land-cover classification application for hyperspectral data with low computational cost and high classification accuracy.
2014 Vol. 27 (3): 248-255 [Abstract] ( 389 ) [HTML 1KB] [ PDF 0KB] ( 101 )
256 Clearance of Flight Control Law Based on Cultural Differential Evolution Algorithm
LI Ai-Jun,WANG Jing,LI Jia,WANG Chang-Qing
Aiming at the slow converge rate in traditional cultural algorithm and lower use efficiency of knowledge about evolutionary information in differential evolution algorithm,a new cultural differential evolution algorithm is proposed. The cultural algorithm is utilized as the framework of the proposed algorithm,in which the evolution in population space consists of mutation,crossover and selection of the differential evolution. In addition,the population space evolution is guided by the belief space knowledge. According to the flying quality specifications,a nonlinear criterion is presented. The proposed algorithm is then applied to evaluate angle of attack limit exceedance criterion,which is current widely used in the aerospace industry. The full authority flight control law of the Aero-Data Model in Research Environment (ADMIRE) is evaluated with uncertainties by the proposed algorithm,which overcomes the limitations of traditional grid-based ones. The simulation results validate that the reliability,computational complexity and efficiency of the proposed algorithm outperform those of the modified differential evolution algorithm,especially in searching for the worst uncertain parameter combinations for the whole flight envelope.
2014 Vol. 27 (3): 256-262 [Abstract] ( 437 ) [HTML 1KB] [ PDF 0KB] ( 127 )
263 Dual Weighted Multi-feature Texture Segmentation Based on QWT and GLCM
LI Ming,LU Fang-Bo,CHEN Hao
Traditional feature weighting algorithms only weight sample globally for mixed attribute data,which ignores the fact that different feature extraction methods are suited to extract different aspects of texture feature. Therefore,a dual weighted strategy is proposed based on the validities and feature importance. Firstly,the fused features of quaternion wavelet transform and gray level cooccurrence matrix are clustered by the k-means algorithm,and the initial cluster centers are regarded as a reference. The k-nearest neighbor samples extracted from each cluster center are regarded as double weighted training sample sets. Then,the problem of weights inside feature is solved by using modified ReliefF algorithm and correlation measure,and the problem of weights between features is solved by using Support Vector Machine. The experimental results show that the proposed method has a good performance in synthetic textures and natural texture images,and has higher segmentation accuracy than other feature weighting algorithms.
2014 Vol. 27 (3): 263-271 [Abstract] ( 406 ) [HTML 1KB] [ PDF 0KB] ( 105 )
272 A Dynamic Guided Multi-objective Optimization Strategy Based on Preference Information
ZHENG Jin-Hua,JIA Yue
The focus of the traditional multi-objective evolutionary algorithms is to obtain the optimal solution set distributed in the entire Pareto frontier. However,in reality problems,the decision makers are merely interested in certain regions of the Pareto frontier. Therefore,it is significant to take the preference information of decision-makers into multi-objective evolutionary algorithms. Thus,how to reduce computing resource and obtain Pareto optimal set effectively in preference regions becomes a hot topic in the research. Aiming at the problem,a dynamic heuristic multi-objective optimization strategy is proposed based on the preference information. The parameter ε is adjusted to reflect the dynamics of the guided regions,and another parameter is set to control the size of preference range of DM. The strategy employs the distance between solution set and the guided regions as a factor for selection strategy. The experimental results show the proposed algorithm with this strategy has a good performance especially on the convergence.
2014 Vol. 27 (3): 272-280 [Abstract] ( 412 ) [HTML 1KB] [ PDF 0KB] ( 63 )
281 null
HAN Su-Qing,YIN Gui-Mei

null
2014 Vol. 27 (3): 281-288 [Abstract] ( 281 ) [HTML 1KB] [ PDF 0KB] ( 125 )
模式识别与人工智能
 

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