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.10, Published 2014-10-30
Papers and Reports    Researches and Applications   
   
Papers and Reports
865 Related Theoretical Analysis of Diversity-Based Semi-supervised Learning
JIANG Zhen, ZHAN Yong-Zhao
Diversity-based semi-supervised learning is the combination of semi-supervised learning and ensemble learning. It is a research focus in machine learning. However, its related theoretical analysis is insufficient, and the presence of distribution noise is not taken into account in these researches. In this paper, according to the characteristic of diversity-based semi-supervised learning, a hybrid classification and distribution (HCAD) noise is defined firstly. Then, probably approximately correct (PAC) analysis for diversity-based semi-supervised learning in the presence of HCAD noise and its application of the theorem are given. Finally, based on the voting margin, an upper bound is developed on the generalization error of multi-classifier systems with theoretic proofs in the presence of HCAD noise. The proposed theorems can be used to design diversity-based semi-supervised learning algorithms and evaluate their generalization ability, and they have a promising application prospect.
2014 Vol. 27 (10): 865-872 [Abstract] ( 481 ) [HTML 1KB] [ PDF 349KB] ( 31063 )
873 The Complementary Equivalence Relation Based Rough Set
MA Zhou-Ming, LI Jin-Jin
A binary relation can be regarded as a set, and thus its complementary set is taken into account. Based on the set attribute of the equivalence relations, the definition of the complementary equivalence relation is given firstly, and its characterization of the relationship among the internal elements is probed. Firstly, the rough set based on the complementary equivalence relation is constructed, and the basis of its axiomatization is discussed as well. Secondly, the mutual relation between the rough set based on the complementary equivalence relation and the classical rough set is studied. And by the proposed rough set, the corresponding calculations of the classical approximation operators is predigested,and the key knowledge like basic accurate set is also fixed.
2014 Vol. 27 (10): 873-878 [Abstract] ( 393 ) [HTML 1KB] [ PDF 308KB] ( 468 )
879 Multi-view Image Registration Method Based on Fuzzy Matching
DONG Tian-Zhen, DENG Ting-Quan, DAI Jia-Shu, XIE Wei , MA Ming-Hua
According to the characteristics of spatial multi-view images, a fuzzy matching based multi-view image registration method is proposed by the principle of coarse-to-fine. Based on image segmentation, the uncertainty of the information from multi-view images is taken into account. The robust regional features such as area, dominant hue and second order moments of brightness are regarded as descriptors of connected regions and then the connected regions are fuzzed. By introducing fuzzy implication, the matching degree between connected regions in multi-view images is calculated. Then, the best matching relation between connected regions is built via fuzzy reasoning. Finally, the feedback correction is used for matching relationship between feature points of connected regions. The adaptive accurate registration between multi-view images is achieved. Meanwhile, the validity of the proposed method is demonstrated through experiments.
2014 Vol. 27 (10): 879-886 [Abstract] ( 463 ) [HTML 1KB] [ PDF 735KB] ( 1020 )
887 Representation Method of Dynamic Uncertain Knowledge Based on Fuzzy-Petri Net and Genetic-PSO Algorithm
PENG Xun, WANG Wei-Ming, GU Chao-Chen, HU Jie
To represent and reason uncertain knowledge in the complex system dynamically and effectively, a self-adaptive method based on fuzzy-Petri net (FPN) and genetic particle swarm optimization (GPSO) algorithm is proposed. In this method, the knowledge-representation model based on FPN is established to build the mathematical model. And the GPSO is used in self-learning of the uncertain parameters to achieve self-adaptation of the model. Finally, a servo mechanism fault diagnose of launch vehicle is used to verify the proposed method.
2014 Vol. 27 (10): 887-894 [Abstract] ( 368 ) [HTML 1KB] [ PDF 618KB] ( 897 )
894 Face Recognition of 2DLDA Based on ULBP Eigensubspace
WU Huang-Peng, DAI Sheng-Kui
The image is segmented at different levels to extract the uniform local binary pattern (ULBP) histogram features of the sub-block images. The global and local features are taken into account, and meanwhile the processing space is converted from the gray space to ULBP feature subspace. Consequently, the correlation between row vectors can be eliminated effectively. Thus, the discriminant projection matrix is performed better through row two-dimensional linear discriminant analysis (R2DLDA). Experimental results on ORL, YALE and FERET databases show that compared with some common methods based on 2DLDA and multilevel LBP, the proposed method achieves a higher recognition rate with a low feature dimension, which proves its effectiveness.
2014 Vol. 27 (10): 894-899 [Abstract] ( 433 ) [HTML 1KB] [ PDF 457KB] ( 1158 )
900 Image Sparse Decomposition Algorithm Based on Multi-population Discrete Differential Evolution
HUANG Ya-Fei, LIANG Xi-Ming, CHEN Yi-Xiong, CHEN Li-Fu
To obtain the sparsest representation of an image using a redundant dictionary is NP-hard, and the existing sub-optimal algorithms for solving this problem such as matching pursuit (MP) are highly complex. An image sparse decomposition algorithm based on multi-population discrete differential evolution for multi-component Gabor dictionaries is proposed. Three sub-populations are adopted to search the best matching atoms in different sub-dictionaries, and the correlation coefficient is used to solve overlap-matching in updating process of residual image. To maintain the population diversity, several mutation operators are employed to generate the offspring population in the proposed algorithm. Experimental results show that the sparse approximation performances of the proposed algorithm are comparable with fast matching pursuit (FMP) algorithm. Meanwhile, the computation speed is improved. The proposed algorithm obtains competitive performance compared with other sparse representation methods based on evolution algorithm. Finally, the rationality of the parameters setting in the proposed algorithm is verified by result analysis.
2014 Vol. 27 (10): 900-906 [Abstract] ( 445 ) [HTML 1KB] [ PDF 618KB] ( 868 )
907 Situation Index Extraction Algorithm Based on Improved Discernibility Matrix and Expert Knowledge
TANG Cheng-Hua, TANG Shen-Sheng, XIE Yi
Aiming at the problems of difficult and inefficient sample index extraction in the general reduction algorithm, rough set theory is introduced to extract situation index. The decision table information is reduced based on the discernibility matrix compression and classified selection, and the index selection is adjusted by combining the importance measure of the export knowledge. Meanwhile, a situation index extraction algorithm based on improved discernibility matrix and expert knowledge is proposed .It is analyzed and verified in the example of situation index system. Experimental results show that the proposed algorithm has good effect on the situation index reduction, and the extracted indexes are rational in the network security assessment. It provides a feasible solution for extracting the situation index.
2014 Vol. 27 (10): 907-914 [Abstract] ( 398 ) [HTML 1KB] [ PDF 404KB] ( 819 )
Researches and Applications
915 Sparse Coding Model Based on Kernel Laplacian for Image Classification
LIU Yue, PENG Hong-Jing, QIAN Su-Jing, SHI Wei
In bag-of-words with sparse coding model, similar features can be encoded as various sparse coding combinations due to the over-completeness of the codebook, which results in totally different visual words. In this paper, a sparse coding method based on kernel Laplacian for image classification is proposed. Firstly, a Laplacian matrix is constructed to capture geometric dependencies between the features in high-dimensional kernel space, and thus the similarity of sparse coding between the similar features can be maximally preserved. Secondly, the objective function is optimized for codebook learning by fixing codebook and sparse matrix alternately, and feature-sign search algorithm is used for sparse coding of the features. Finally, the one-to-all linear SVM classifier is applied to classify images. The experimental results on several datasets show the proposed algorithm decreases the quantization error dramatically and improves the classification performance.
2014 Vol. 27 (10): 915-920 [Abstract] ( 537 ) [HTML 1KB] [ PDF 355KB] ( 985 )
921 Variable Influence Space Based Uniformity Metric Method for Solution Sets of Multi-objective Evolutionary Algorithms
ZHENG Jin-Hua, HUANG Duan, WANG Kang, ZHANG Zuo-Feng
Uniformity Evaluation of solutions is one of the most important issues of performance assessment. The views of facing individuals and facing space are combined to construct a variable influence space of solutions. A variable influence space based uniformity metric method for solution sets of multi-objective evolutionary algorithms is proposed in this paper. The metric can be used to compare the performance of different multi-objective optimization techniques by evaluating the relative degree of uniformity of a solution set in the influence space. Experimental results show the feasibility and effectiveness of the proposed metric.
2014 Vol. 27 (10): 921-929 [Abstract] ( 393 ) [HTML 1KB] [ PDF 777KB] ( 1070 )
930 A Payoff Distribution Strategy for Overlapping Coalitions Based on Bargaining
ZHANG Guo-Fu, ZHOU Peng, SU Zhao-Pin, YANG Ren-Zhi, JIANG Jian-Guo
In multi-agent system (MAS), payoff distribution for overlapping coalitions is a difficult problem in overlapping coalition formation (OCF). In this paper, the possible resource conflicts in OCF are discussed firstly, then some important characteristics of the OCF model are deduced. Based on those results, the strategy of bargaining is introduced to allocate tasks to agents in coalitions, and the payoff of coalitions is distributed according to the principle of non-reducing utility. Finally, the analysis of a specific example shows the feasibility of the proposed method.
2014 Vol. 27 (10): 930-938 [Abstract] ( 453 ) [HTML 1KB] [ PDF 453KB] ( 763 )
939 Real-Time System for Human Detection and Tracking at Different Distances
YUAN Yang, HUANG Di, WANG Yun-Hong
A real-time robust human detection and tracking system is proposed, which can detect people in the monitoring area and then keep tracking. To reduce the working range, a background subtraction technique is used to segment the moving foreground and the background. Since each body feature has its optimum working distance, several different detectors such as frontal face, head, and pedestrian are combined. By taking the video sequence continuity and the human body geometry constraint into account, robust real-time detection is achieved. The proposed system reduces the occurrence of tracking failure and enhances performance even with dramatic distance change between camera and people.
2014 Vol. 27 (10): 939-945 [Abstract] ( 407 ) [HTML 1KB] [ PDF 1063KB] ( 2163 )
946 Bilinear Regression Based Expression Face Neutralization Preserving Person-Specific Characters
CHEN Ying, ZHANG Long-Yuan, YI Xiao-Bin
In traditional learning-based expression neutralization methods, person-specific characters are removed from the visual neutral face.To solve this problem, an expression neutralization algorithm based on bilinear kernel rank reduced regression (BKRRR) is proposed. The BKRRR algorithm is designed to synthesize virtual expression and virtual neutral images from training samples simultaneously. An expression mask is established using differences of the two images. The test expression image is warped to neutral template by piece-wise affine warp. An image fusion strategy based on Poisson equation is then designed. The visual BKRRR neutral image is used as foreground, the warped image as background and the expression mask as foreground mask. And thus the visual neutralized face image with person-specific characters preserved is obtained. Experimental results show that the synthesized visual neutral image outperforms other learning-based methods in both visual and objective evaluation, and the accuracy of expression invariant face recognition is improved.
2014 Vol. 27 (10): 946-953 [Abstract] ( 437 ) [HTML 1KB] [ PDF 1289KB] ( 1547 )
954 Face Recognition Using Sparse Coding by Embedding Maximum Block Similarity
WANG Shu-Xian, XIONG Cheng-Yi, GAO Zhi-Rong, ZHOU Cheng, HOU Jian-Hua

The performance of sparse representation based face recognition (SRFC) can be effectively improved by embedding a priori of similarity information. Aiming at expressions variations, partial occlusions and disguise in the uncontrolled face images, SRFC by embedding maximum block similarity information is proposed. Firstly, the training samples and query samples are divided into multiple non-overlapping blocks in the same way. Secondly, the similarities of corresponding blocks between the query samples and the training samples are calculated. Then, the maximum value is extracted to measure the similarity of inter-images. Finally, the extracted maximum block similarity information is embedded into sparse representation stage. Experimental results on AR face databases show that the proposed method achieves better recognition performance compared with those based on embedding global similarity, especially when both training images and query images contain expression, occlusions or disguise.
2014 Vol. 27 (10): 954-960 [Abstract] ( 396 ) [HTML 1KB] [ PDF 534KB] ( 721 )
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