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.11, Published 2014-11-30
Papers and Reports    Researches and Applications   
   
Papers and Reports
961 A Divide-and-Conquer Based Multidimensional Scaling Algorithm
QU Tai-Guo, CAI Zi-Xing
As a typical method for multivariate statistical analysis, multidimensional scaling(MDS) has been widely used in dimensionality reduction and visualization. Based on a distance matrix of n samples, the coordinates of MDS in a low-dimensional Euclidean space are obtained. The time complexity for classical MDS algorithm(CMDS) is Θ(n3), which affects the speed of MDS. A new MDS algorithm based on the idea of divide-and-conquer is proposed. The distance matrix is divided into several submatrices along its diagonal, then the submatrices are solved respectively. With orthogonal transformation and translation transformation, all of these subproblem solutions are aligned to get the global solution to the original distance matrix. The result of the proposed algorithm is completely the same as that of CMDS. When the dimensionality of sample space is far less than the number of samples, the time complexity of the proposed algorithm is only Θ(nlgn). Thus, compared with CMDS, the speed of the proposed algorithm is greatly improved, which makes MDS applicable to larger datasets.
2014 Vol. 27 (11): 961-969 [Abstract] ( 414 ) [HTML 1KB] [ PDF 639KB] ( 1037 )
970 Study on Self-Decision of Cyber-Physical Systems Based on Similarity Computation
ZHOU Wang-Ping, WANG Guo-Dong
Cyber-Physical systems (CPS) are getting more and more popular. How to make the system automatically capture the changes of complex statuses and take proper actions responding to the changes is one of the key problems of CPS. Combining with the reinforcement learning algorithm, a novel self-decision algorithm of CPS based on similarity computation, called Similarity Computation Based on Reinforcement Learning Algorithm(SCBRLA), is proposed to solve this problem. In this algorithm, the features of both system and system targets are firstly extracted, and then the similarities of current system state and target states are computed. Based on the computation result, system takes corresponding actions and decides the execution order of those actions. The proposed algorithm can be well used to analyze strategies of system self-decision when it receives attacks. The simulation results show that the proposed algorithm can help systems realize self-decision, and it has faster response speed compared with traditional method.
2014 Vol. 27 (11): 970-976 [Abstract] ( 351 ) [HTML 1KB] [ PDF 439KB] ( 632 )
977 PK-SVD Filter for Impulse Noise Based on Non-noisy Pixel Reconstruction
HUANG Yan-Wei, QI Bing-Lu
An improved K-SVD method based on non-noisy pixel reconstruction (PK-SVD) is proposed to filter impulse noise. In the phase of image reconstruction, non-noisy pixels are applied in the construction of optimal function to obtain the reconstructed image and improve the filtering performance, and the optimal function is solved by integrating the hierarchical property into the OMP algorithm. In the phase of dictionary training, PK-SVD uses the iterant K-singular value decomposition to renovate both atoms and their coefficients rather than fixes the coefficients. The simulation results show that compared with the other three methods, PK-SVD obtains the sparsest dictionary and the clearest image with higher peak signal to noise ratio.
2014 Vol. 27 (11): 977-984 [Abstract] ( 382 ) [HTML 1KB] [ PDF 1449KB] ( 688 )
985 Pest Image Recognition of Multi-feature Fusion Based on Sparse Representation
HU Yong-Qiang, SONG Liang-Tu, ZHANG Jie, XIE Cheng-Jun, LI Rui
Aiming at the characteristics of different pest images with different colors, shapes and textures, a pest recognition method based on sparse representation and multi-feature fusion is proposed, which uses a matrix of labeled training samples to construct different dictionaries. The recognition result is achieved by solving optimal sparse coefficients with the corresponding feature dictionary. Furthermore, a novel learning method, which can be improved efficiently by jointly optimizing classifier weights, is presented to effectively fuse multiple features for pest categorization. The experimental results on real datasets show that the proposed method performs well on pest species recognition either in laboratory or in farmland.
2014 Vol. 27 (11): 985-992 [Abstract] ( 523 ) [HTML 1KB] [ PDF 619KB] ( 1041 )
993 An Adaptive Multi-threshold Image Segmentation Method without Preassigning the Number of Segmented Regions
CHEN Ya-Jun, LIU Ding, LIANG Jun-Li, ZHANG Xin-Yu
To solve the problem that it is difficult to choose the number of segmentation regions for multi-threshold image segmentation, an adaptive multi-threshold image segmentation method based on Reversible Jump Markov Chain Monte Carlo (RJMCMC) method is proposed. Histogram-based image segmentation is essential to search the bottom between peaks. However, the multi-threshold segmentation number is difficult to determine and not all local peaks follow Gaussian distribution. Therefore, mixture of α-stable distributions is adopted to fit image gray level histogram. Firstly, a hierarchical Bayesian probability model is established with the number of local peaks and the various parameters for each component. Then, posterior probability reasoning based on RJMCMC is implemented to adaptively obtain the best number of α-stable distribution function and the parameters for each distribution. The experimental results on the single crystal pulling image, the simulated magnetic resonance imaging (MRI) image and international standard test images show that the image segmentation model is accurately constructed by the proposed method, and multi-threshold segmentation results of images are satisfactory.
2014 Vol. 27 (11): 993-1004 [Abstract] ( 410 ) [HTML 1KB] [ PDF 1631KB] ( 647 )
1005 Lifecycle-Based Binary Ant Colony Optimization Algorithm
CHENG Mei-Ying, NI Zhi-Wei, ZHU Xu-Hui
The biological life cycle in natural ecosystem is introduced into binary ant colony optimization algorithm, and the main idea is to execute breeding, migrating and dying operations by setting relevant nutritious threshold value to the ants. Thus, the dynamic diversity of the population is maintained and the drawback that binary ant colony optimization algorithm easily traps in local optimum is overcome. The proposed algorithm, lifecycle-based binary ant colony optimization algorithm (LCBBACO), is combined with fractal dimension to attribute reduction problem. The experimental results on 6 UCI datasets show that the method has preferable feasibility and effectiveness.
2014 Vol. 27 (11): 1005-1014 [Abstract] ( 468 ) [HTML 1KB] [ PDF 533KB] ( 594 )
1015 Analysis of General Model and Classical Algorithms for Spectral Clustering
GUAN Tao, YANG Ting
Spectral clustering is able to find the nonlinear low-rank structure of data, and it is widely applied to pattern recognition.Besides,spectral clustering has some internal relations with graph models, manifold embedding and integral operator theory from the theoretical view.However, it is lack of systematically theoretical research in these aspects. The general model of spectral clustering is introduced from the latest research outcomes, that is, eigenfunctions learning of integral operators inreproducing kernel Hilbert space(RKHS). Subsequently, the internal relations of spectral clustering with KPCA, kernel k-means,Laplacian eigenmap, manifold learning, and discriminant analysis are discussed. Then, some classical spectral clustering algorithms are introduced, such as NJW algorithm, Ncut, spectral clustering based on Nystrm method, multiscale spectral clustering algorithm. At last, trends and possible difficulties in spectral clustering are summarized.
2014 Vol. 27 (11): 1015-1025 [Abstract] ( 448 ) [HTML 1KB] [ PDF 1159KB] ( 1143 )
Researches and Applications
1026 Variable Step-Size Nonholonomic Natural Gradient Algorithm Based on Sign Operator
JI Ce, YANG Kun, WANG Yan-Ru, LIU Meng-Die
By introducing the nonholonomic constraints, the nonholonomic natural gradient algorithm effectively overcomes the shortcoming and the insufficiency of the traditional natural gradient algorithm, namely,it can still work well when the amplitude of source signal changes rapidly over time or is equal to zero in a certain period of time. Meanwhile, the sign operator derived from a general dynamic separation model can improve the convergence of the algorithm. Thus, a nonholonomic natural gradient algorithm based on the sign operation is obtained by combining the above two ideas. Furthermore, a variable step-size based on the gradient of cost function is also applied to the proposed algorithm to balance the contradiction between the convergence speed and the steady-state error. The simulation results show that the performance of the proposed algorithm is superior to that of traditional algorithm, and it improves convergence speed without worsening the steady-state error seriously.
2014 Vol. 27 (11): 1026-1031 [Abstract] ( 371 ) [HTML 1KB] [ PDF 1118KB] ( 887 )
1032 Fuzzy Neighborhood Preserving Embedding Algorithm Based on Common Vector
ZHENG Hai-Tao, ZHENG Gang-Min, MA Xiao-Hu
Neighborhood preserving embedding directly reconstructs the sample by its K-nearest neighbors. However, it does not distinguish the importance between intra-class neighbors and inter-class neighbors, which leads to poor recognition performance. In this paper, a common vector-based fuzzy neighborhood preserving embedding (FNPE/CV) algorithm is proposed.Firstly, the degree of membership of every sample for each class is obtained based on the class labels of its K-nearest neighbors. Then, every sample is reconstructed by the common vector and its membership grade for every class. Finally, the problem of minimizing the residual between original sample and its reconstruction sample is converted to solve the generalized eigenvalue problem to obtain the final projection transformation matrix. After the projecting, FNPE/CV minimizes the difference among intra-class samples and separates inter-class samples as far as possible. The experiments on ORL, Yale, AR and PIEC29 face databases demonstrate the effectiveness of the proposed algorithm.
2014 Vol. 27 (11): 1032-1039 [Abstract] ( 399 ) [HTML 1KB] [ PDF 1136KB] ( 513 )
1040 Seam Carving Based Image Resizing with New Seam Energy Function
NIE Dong-Dong, MA Qin-Yong
The image quality of seam carving based image resizing method heavily depends on the definition of seam energy function and the result of seam extraction. A seam energy function is proposed to calculate the visual perception importance of each seam, in which an incremental energy term is defined to measure the local image aliasing after seam pixels are deleted, and a maximum pixel energy term is introduced to reduce the seam energy of random texture regions. The experimental results show that compared with several other classical methods, the proposed energy function have more reasonable location, its results are more similar to the original images, and it has fewer aliasing problems in the details of images key objects.
2014 Vol. 27 (11): 1040-1046 [Abstract] ( 355 ) [HTML 1KB] [ PDF 1003KB] ( 541 )
1047 Evaluation of Emergency Disposal Schemes Based on Cloud Model and Fuzzy Aggregation
SU Zhao-Pin, ZHANG Ting, ZHANG Guo-Fu, YOU Xiao-Quan, JIANG Jian-Guo

The evaluation of emergency disposal schemes is a key topic in disaster emergency response. However, most existing methods only depend on the commander's personal knowledge and experience, which cannot thoroughly consider suggestions of experts in different fields. To tackle such situation, cloud model and fuzzy soft sets are introduced to evaluate emergency disposal schemes. Firstly, cloud model is used to transform qualitative variables into quantitative variables for evaluation information given by experts in different fields. In addition, fuzzy soft sets are adopted to aggregate all quantitative evaluation information to achieve comprehensive evaluation. Finally, the experimental results show that the proposed method is feasible and can provide a theoretical and technical guidance for practical emergency decision making.
2014 Vol. 27 (11): 1047-1056 [Abstract] ( 371 ) [HTML 1KB] [ PDF 765KB] ( 746 )
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