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.5, Published 2014-05-30
Papers and Reports    Researches and Applications   
   
Papers and Reports
385 Research on the Properties of Orthogonal Polynomial Kernel Functions
TIAN Meng ,WANG Wen-Jian
The choice of kernel function and its parameters a core problem of support vector machine (SVM). Based on orthogonality and variability of orthogonal polynomial functions, kernel functions are constructed to be used as general kernel functions instead of some common kernels, such as polynomial kernel and Gaussian kernel. Generally, the kernel parameters are chosen only from natural number, which facilitates the kernel parameter tuning. 6 sets of orthogonal polynomial kernel functions based on Chebyshev polynomial, Legendre polynomial, Hermite polynomial, and Laguerre polynomial are discussed. The properties of these kernel functions are studied, and their robustness and generalization performance on some test datasets are compared. The obtained results provides theoretical basis and technical support for SVM classification.
2014 Vol. 27 (5): 385-393 [Abstract] ( 536 ) [HTML 1KB] [ PDF 1262KB] ( 1038 )
394 Linguistic-Valued Intuitionistic Fuzzy 2-Tuple Representation Model
ZOU Li,ZHANG Yun-Xia,Gao Wei
A kind of linguistic-valued intuitionistic fuzzy 2-tuple representation model is proposed based on the linguistic truth-valued intuitionistic fuzzy lattice. The operations and the properties of linguistic-valued intuitionistic fuzzy pair are discussed. Based on the proposed model, a kind of linguistic-valued intuitionistic fuzzy aggregation approach is defined. Arithmetic mean aggregation operator and weighted average aggregation operator are given respectively. Using the aggregation operations of the linguistic-valued intuitionistic fuzzy pairs, the information loss is reduced in the process of the aggregation for linguistic-valued intuitionisttic fuzzy pair. A recognition process of disease diagnosis is given to illustrate that the proposed approach is more effective for pattern recognition under a fuzzy environment with both positive evidence and negative evidence.
2014 Vol. 27 (5): 394-402 [Abstract] ( 454 ) [HTML 1KB] [ PDF 478KB] ( 478 )
403 EEG De-noising Method Based on Double-Density Discrete Wavelet Transform Using Neighbor-Dependency Thresholding
LUO Zhi-Zeng,ZHOU Ying,GAO Yun-Yuan
To eliminate the noise mixed in Electroencephalogram (EEG), an EEG de-noising method is proposed based on double-density discrete wavelet transform using neighbor-dependency thresholding. Firstly, high frequency coefficients of multilayer signals are obtained by double-density discrete wavelet decomposition. Then, the wavelet coefficients are shrunk with neighbor-dependency thresholding algorithm, which takes the statistical dependencies of the wavelet coefficients into account. Finally, the de-noising signal is obtained by reconstructing shrunk wavelet coefficients. The simulation results of the de-noising experiments on standard noise-adding signal and real EEG show that compared to the first generation discrete wavelet algorithm and traditional soft threshold methods, the proposed de-noising algorithm has the benefits of higher SNR, lower RMSE and Errmax.
2014 Vol. 27 (5): 403-409 [Abstract] ( 344 ) [HTML 1KB] [ PDF 876KB] ( 546 )
410 A Hierarchical Clustering Algorithm Based on Asymmetric Distance
HAN Zhong-Ming,CHEN Ni,ZHANG Hui,YANG Wei-Jie
Hierarchical clustering algorithm is applied in many research fields such as data mining and machine learning. Most existing hierarchical clustering algorithms are dependent on symmetrical distances definition. In this paper, a hierarchical clustering algorithm is proposed based on asymmetric distance. With respect to asymmetric distance characteristics, a selective factor and corresponding calculation formula are proposed. The single linkage, full linkage and average linkage algorithms for the asymmetric hierarchical clustering algorithm are implemented. The hot tags from main social bookmarking systems are extracted and an asymmetric distance is defined based on co-occurrence frequency of different tags. The experimental results show that the proposed algorithm outperforms the clustering algorithm based on symmetrical distance. The cophenetic coefficient is also used to evaluate effectiveness of the algorithm.
2014 Vol. 27 (5): 410-416 [Abstract] ( 449 ) [HTML 1KB] [ PDF 1260KB] ( 20104 )
417 Collaborative Filteration Recommendation Algorithm Based on Trust Computation
DU Yong-Ping,HUANG Liang,HE Ming
Collaborative filteration is one of the most widely used recommendation strategies, in which data sparsity problem and expansion difficulty exist. Based on traditional user-based collaborative filtering algorithms, the trust computation is introduced into the process of recommendation. Making full use of the propagation characteristics of trust relationship under some conditions, a hybrid network composed of the user reputation-trust and the user local-trust is designed and built. And the user rating similarity is combined with trust evaluation of the hybrid network, which helps users to discover more two-dimensional similarity neighbors based on trust and interest factors. The proposed method is validated by the experiment on Epinions dataset with Mean Absolute Error (MAE) and Root Mean Square Error (RSME) as the evaluation index. The results show that compared to the traditional collaborative filtering recommendation algorithms, MAE of the proposed method increases about 6.8% and the optimal value reaches 0.7513, and the t-test results also show that the proposed method improves the performance significantly.
2014 Vol. 27 (5): 417-425 [Abstract] ( 389 ) [HTML 1KB] [ PDF 873KB] ( 721 )
426 Discriminative Rival Penalization Controlled Competitive Learning Algorithm
ZHANG Feng,ZHAO Jie-Yu,ZHU Shao-Jun
Competitive learning is an important approach for clustering analysis. The rival penalized competitive learning (RPCL) algorithm has the ability of selecting the correct number of clusters automatically, but its performance is sensitive to the selection of learning rate and de-learning rate. In fact, it is unreasonable that all the rival units are treated as redundant units to be penalized in the variant algorithm called rival penalization controlled competitive learning (RPCCL). In this paper, a discriminative rival penalization controlled competitive learning (DRPCCL) is presented. The learning rate of winningunits adaptively adjusts during iteration in the proposed method. Meanwhile, a discriminative penalization controlled mechanism is used to discriminate the redundant units and the correct units in the rival units. The correct units and redundant units are given a slight penalization and a heavier penalization respectively, which makes this algorithm get exact number of clusters and reasonable centre of clusters. The experimental result demonstrates that compared with RPCL and RPCCL, DRPCCL achieves more accurate performance.
2014 Vol. 27 (5): 426-434 [Abstract] ( 315 ) [HTML 1KB] [ PDF 763KB] ( 620 )
Researches and Applications
435 A Vehicle Recognition Method Based on Kernel K-SVD and Sparse Representation
SUN Rui,WANG Jing-Jing
The classification and recognition of vehicle is of great importance in the research of intelligent transportation system. A method based on PCA, kernel K-SVD and sparse representation classification method is proposed for two-class supervised classification. Firstly, PCA is used in this method to train both vehicle and non-vehicle images for feature extraction and dimensionality reduction. Then, the Gaussian-Kernel function is used to map the matrix to the high-dimensional space, and K-SVD is applied to train the high-dimension feature matrix for two corresponding dictionaries in this space. Finally, training images based on l1-minimization sparse coefficient are used to linearly represent test images. The experiments are carried out and the results show that the performance of the proposed method on the partially-covered case is obviously better than that of other classical methods.
2014 Vol. 27 (5): 435-442 [Abstract] ( 610 ) [HTML 1KB] [ PDF 924KB] ( 938 )
443 Online Clustering of Evolution Data Stream Based on Affinity Propagation Clustering
ZHANG Jian-Peng,CHEN Fu-Cai,LI Shao-Mei,Liu Li-Xiong
To improve the accuracy and timeliness of data stream clustering, an efficient data stream clustering algorithm is proposed with temporal characteristics and affinity propagation methods (TCAPStream).The algorithm merges the newly detected class mode into clustering model by using the improved WAP algorithm, meanwhile, the temporal evolution characteristic of the data stream is reflected by using micro-cluster temporal density. Besides, the online dynamic deletion mechanism is proposed to maintain the micro-clusters. It makes the algorithm model reflecting both temporal and distribution characteristics of data stream to obtain more accurate clustering results. The experimental results show that the proposed algorithm not only has good clustering effect in several artificial datasets and real datasets, but also has good flexibility and extensibility.
2014 Vol. 27 (5): 443-451 [Abstract] ( 354 ) [HTML 1KB] [ PDF 805KB] ( 724 )
452 Eigenvector Selection Algorithm for Spectral Clustering Based on Dynamic Selective Ensemble
WANG Xing-Liang,WANG Li-Hong,WU Shuan-Hu
Since the corresponding eigenvectors of k maximum eigenvalues do not always achieve the optimal clustering results, the clustering performance is improved by selective integrated approach for eigenvector groups involving the selection of base eigenvector group and selective integration strategy. Constraint score is used to evaluate eigenvectors by the pair-wise constraint information of training data, and some preferable base eigenvector groups are obtained. For each testing data, the clustering accuracy of l-nearest neighbors from training dataset are used to dynamically evaluate eigenvector groups, and several accurate eigenvector groups are selected to vote. To test the obtained eigenvector groups, spectral clustering is carried out on the corresponding eigenvectors of testing dataset. The clustering results are aligned and the final experimental results are obtained. The experimental results on UCI benchmark datasets show that the proposed algorithm improves the clustering performance of testing data.
2014 Vol. 27 (5): 452-462 [Abstract] ( 346 ) [HTML 1KB] [ PDF 624KB] ( 478 )
463 A Multi-class Feature Selection Algorithm Based on Support Vector Machine
DAI Kun,YU Hong-Yi,LI Qing
Most existing feature selection algorithms usually select only one feature randomly from the highly correlated feature subset with great contribution to classification,which results in the degradation of data readability and classification performance. To overcome the problem, a multi-class feature selection algorithm based on support vector machine(MFSSVM)is proposed. The proposed feature selection algorithm permits highly correlated features to be selected or removed together, and it allows dimension reduction while obtaining effective features. The experimental results on both simulated datasets and benchmark datasets illustrate the feasibility and effectiveness of the feature set selected by MFSSVM.
2014 Vol. 27 (5): 463-471 [Abstract] ( 411 ) [HTML 1KB] [ PDF 1008KB] ( 637 )
472 Immune Task Allocation for Multi-robot System Based on Adjustment Mechanism of Thymic Peptide
YUAN Ming-Xin,LI Ping-Zheng,JIANG Ya-Feng, WANG Sun-An

To solve the task allocation in multi-robot system, a thymic peptide based immune task allocation algorithm (TPITAA) is proposed. Inspired by the mechanism of idiotypic network hypothesis, an immune allocation model is constructed according to the stimulation and suppression among the antigen and antibodies. The robot, robot behavior and task are taken as B cell, antibody and antigen respectively. To further improve the allocation efficiency, a thymic peptide feedback function based on movement direction of robot is defined according to the immune adjustment mechanism of thymic peptide, which realizes the self-adjustment of antibody stimulation level and antibody concentration. The simulation results show that the proposed algorithm realizes the automatic allocation for tasks, reduces the completion time, improves the operating efficiency and solves the cooperation handling well in multi-robot system.

Guide:null
2014 Vol. 27 (5): 472-480 [Abstract] ( 249 ) [HTML 1KB] [ PDF 765KB] ( 406 )
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