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.4, Published 2013-04-30 Orignal Article        Orignal Article 321 Target Position Estimation Aided Swarm Robotic Search under Conditions of Absolute Localization Mechanism ZAN Yun-Long,XUE Song-Dong,ZENG Jian-Chao When swarm robots are controlled cooperatively following the extended particle swarm optimization model for target search,each member robot is guided by social experience and its cognition to move step by step toward the potential goal. However,in this mechanism the social experience must be elected from the cognition of member robots rather than formal fusion. Wireless sensor network is a characteristic of swarm robotic system in aspect of physical properties,on the basis of that,a target position estimation method is employed. The range-based localization of wireless sensor network is used for target position estimation to replace social experience in the existing model. Results of simulations show that the proposed mode of social experience producing outperforms the existing ones when swarm scale is big enough,which makes cooperation search efficiency raised. 2013 Vol. 26 (4): 321-327 [Abstract] ( 702 ) [HTML 0KB] [ PDF 511KB] ( 713 ) 328 Multilevel and Mean Shift Based Image Segmentation Using Kway-Ncut TAN Le-Yi,WANG Shou-Jue A fast image segmentation algorithm is presented,which can segment large images effectively. The kway-normalized cut(Kway-Ncut) graph partitioning is used as a framework of image segmentation. Firstly,the image is pre-segmented by Mean Shift algorithm. Secondly,both the original image and the pre-segment result are compressed into small scale to achieve acceleration. Thirdly,the pairwise pixel similarity is computed in the low-scale image incorporating the prior knowledge of the pre-segment result and the spatial coherence of pixel. Next,Kway-Ncut is used to partition the graph. Finally,the original pre-segment result is used to recover the details and the boundaries of the segmentation. Besides,the recover method is explained through Bayes rules. The proposed algorithm is applied to segment static images and the results show that the proposed method outperforms other ones due to its lower computational complexity and great accuracy. 2013 Vol. 26 (4): 328-336 [Abstract] ( 736 ) [HTML 0KB] [ PDF 7617KB] ( 685 ) 337 Web Services Trust Computation Based on Social Network Dynamic Feedback ZHANG Pei-Yun,CHEN En-Hong,LI Bo In the current service computing background,it is difficult for users to obtain the trusted services. Aiming at this problem,a Web services trust model based on social network dynamic feedback is proposed. An algorithm for computing direct trust value of the service is designed by using individual users experience. The transactions of the services are dynamically tracked and monitored by the proposed algorithm. When a user has no direct experience of a service,based on the users trust values in the social network the indirect trust value of the service can be calculated by gathering a number of other users direct trust values of the service. As the social network nodes may recommend untrue services,the correction factor is used to amend the trust of social network peers. The analysis of the proposed algorithms shows that this method is feasible and effective. 2013 Vol. 26 (4): 337-343 [Abstract] ( 587 ) [HTML 0KB] [ PDF 522KB] ( 629 ) 344 Fractal Evolutionary Particle Swarm Optimization QIU Xiao-Hong,QIU Xiao-Hui,GONG Yao-Teng Based on the classic particle swarm optimization (PSO) algorithm,a fractal evolutionary particle swarm optimization(FEPSO)is proposed . In FEPSO,the charactristic of the irregular motion of fractal Brownian motion model is used to simulate the optimization process varying in unknown mode,and its implied trend part is applied to simulate the optimization index of the global objective function optimum change. Therefore,the individual evolution process is prevented from going too randomly and precociously. Compared with the classic PSO algorithm,a fractal evolutionary phase is included for each particle in FEPSO. In this phase,each particle simulates a fractal Brownian motion with different Hurst parameter to search the solution in sub dimensional space,and its corresponding sub position is updated. The results of simulation experiments show that the proposed algorithm has a robust global search ability for most standard composite test functions and its optimization ability performs better than the recently proposed improved algorithm based on PSO. 2013 Vol. 26 (4): 344-350 [Abstract] ( 621 ) [HTML 0KB] [ PDF 687KB] ( 739 ) 351 One-Class Classifier Algorithm Based on Ensemble Multi-Spanning Trees by Pruning Random Subspace Method HU Zheng-Ping,LIU Kai Due to the redundancy and the noise in high-dimensional data,a covering model constructed from these data can not reflect their distribution information,which leads to the performance degradation of one-class classifiers. To solve this problem,a pruning random subspace ensemble multi-spanning tree method is proposed. Firstly,several random subspaces are created,and minimum spanning tree covering models are constructed in each subspace respectively. Next,pruning ensembles are applied to each classifier by using an evaluation criterion. Finally,these subspace classifiers are integrated into an ensemble classifier by mean combining. Experimental results show that the proposed covering classifier by ensemble multi-trees has a better correct rate in classification than other direct covering classifiers and bagging algorithm. 2013 Vol. 26 (4): 351-356 [Abstract] ( 446 ) [HTML 0KB] [ PDF 388KB] ( 822 ) 357 Degradation Data-Driven Remaining Useful Life Estimation Approach under Collaboration between Bayesian Updating and EM Algorithm SI Xiao-Sheng,HU Chang-Hua,LI Juan,CHEN Mao-Yin Remaining useful life (RUL) estimation is one of the key issues in condition-based maintenance and prognostics and health management. To achieve degradation modeling and RUL estimation for the individual equipment in service,a degradation data-driven RUL estimation approach under the collaboration between Bayesian updating and expectation maximization (EM) algorithm is presented. Firstly,an exponential-like degradation model is utilized to describe the equipment degradation process and the stochastic parameters in the model are updated by Bayesian approach. Based on the Bayesian updating results,the probability distribution of the RUL is derived and the point estimation of the RUL is obtained accordingly. Secondly,based on the monitored degradation data to date,a parameter estimation approach for other non-stochastic parameters in the established degradation model is proved. Furthermore,it is proved that the obtained estimation in each iteration is unique and optimal. Finally,a numerical example and a practical case study are provided to show that the presented approach effectively models degradation process for the individual equipment,achieves RUL estimation,estimates the model parameters and generates better results than a previously reported approach in the literature. 2013 Vol. 26 (4): 357-365 [Abstract] ( 768 ) [HTML 0KB] [ PDF 778KB] ( 1006 ) 366 Stochastic Optimization Based Fast Learning Method on Large-Scale Noisy Datasets WANG Jia-Bao Aiming at large-scale machine learning problems with noise and interference data,the non-convex Ramp loss function is adopted to suppress the influences of noise and interference data,and a fast learning method is proposed for solving the non-convex linear support vector machines based on stochastic optimization. It effectively improves the training speed and the prediction accuracy. The experimental results manifest that the proposed method greatly reduces the learning time,and on the MNIST dataset the training time is reduced by 4 orders of magnitude compared to the traditional learning method. Meanwhile,it improves the prediction speed in a sense and greatly enhances the generalization performance of the classifiers for noisy dataset. 2013 Vol. 26 (4): 366-373 [Abstract] ( 614 ) [HTML 0KB] [ PDF 698KB] ( 805 ) 374 Memory Cell Pruning and Nonlinear Resource Allocation BasedArtificial Immune Recognition System DENG Ze-Lin ,TAN Guan-Zheng,HE Pei To reduce the memory cells of artificial immune recognition system (AIRS) and improve AIRS classification performance,a memory cell pruning and nonlinear resource allocation based artificial immune recognition system (PNAIRS) is proposed. Attribute discretization pre-processing is adopted to compress the training space. Memory cell pruning operation is employed to eliminate the memory cells of low fitness scores,and nonlinear resource allocation is utilized to optimize the classifier. PNAIRS is applied to 6 UCI datasets classification,the classification performance is compared with other classifiers. PNAIRS generates small memory cell population and reaches high classification accuracy,and the classification is finished quickly. The results show that PNAIRS is a high-performance classifier,and it has potential application. 2013 Vol. 26 (4): 374-381 [Abstract] ( 315 ) [HTML 0KB] [ PDF 527KB] ( 599 ) 382 A Local Gibbs Sampling Automatic Inference Algorithm Based on Structural Analysis WANG Hao,CAO Long-Yu ,YAO Hong-Liang ,LI Jun-Zhao In this paper,a local Gibbs sampling inference algorithm of Bayesian networks (S-LGSI) is proposed. Firstly,the S-LGSI algorithm precisely decomposes Bayesian networks based on the analytic idea of junction tree algorithm. Secondly,the suitable local model is chosen by the query node and the evidence node. Then,Gibbs sampling inference algorithm for local network model is utilized. Compared with other current approximate sampling algorithms,the S-LGSI algorithm significantly reduces the calculation dimension. The sampling inference in the local model avoids the statistics of joint sample series and greatly reduces the calculation dimension. The proposed algorithm guarantees the inference precision,as the local model contains important information about the query node. Algorithm analysis and experimental results on Alarm network show S-LGSI significantly reduces the complexity and improves the inference precision. The proposed algorithm has strong practicability,because the inference results of S-LGSI algorithm are basically consistent with the real situation on the Shanghai Stock Exchange network. 2013 Vol. 26 (4): 382-391 [Abstract] ( 547 ) [HTML 0KB] [ PDF 639KB] ( 758 ) 392 Adaptive Kernel Feature Subspace Method for Efficient Feature Extraction ZHANG Zhao-Yang,TIAN Zheng Kernel principal component analysis(KPCA) can extract nonlinear features of datasets. However,its efficiency is inversely proportional to the size of the training sample set. In this paper,an adaptive kernel feature subspace method is proposed to extract features efficiently. This method is methodologically consistent with KPCA,and it improves the efficiency by adaptively selecting the spanning vectors of the KPCA without losing accuracy. Experimental results on two-dimensional data and MNIST datasets show that the proposed method is better than the one associated with KPCA and reference methods. 2013 Vol. 26 (4): 392-401 [Abstract] ( 745 ) [HTML 0KB] [ PDF 1018KB] ( 641 ) 402 Immunization Strategy with Particle Swarm Optimization for Virus Spread Control in Weighted Scale-Free Networks GUO Wen-Zhong,CHEN Guo-Long,WANG Ning-Ning,LIN Bing An immunization strategy based on particle swarm optimization is presented to effectively solve the control of the virus spread in the weighted scale-free networks. Motivated by the idea of network partition,two optimization goals with the sub-network scale and the sum of the sub-network strength are simultaneously taken into account. Moreover,the mutation and the crossover operator of genetic algorithm are introduced to improve the population diversity and avoid the algorithm falling into a local optimal solution prematurely. Simulation results show that the proposed immunization strategy has better performance than the targeted immunization strategy which is generally considered to be highly efficient at present. Through immunizing the specified number of nodes,the network can be well divided into the sub-networks whose nodes number and sum of nodes strength are as small as possible. 2013 Vol. 26 (4): 402-408 [Abstract] ( 353 ) [HTML 0KB] [ PDF 516KB] ( 601 ) 409 Intensity Order Based Mean-Standard Deviation Descriptor WANG Zhi-Heng,ZHI Shan-Shan,LIU Hong-Min Based on mean-standard deviation curve descriptor (MSCD),the intensity order mean-standard deviation descriptor (IOMSD) is proposed by introducing the idea of intensity order partition. Different from the fixed-position partitioning used in the construction of traditional descriptors,the sub-regions are partitioned according to the relationship of intensities among sample points. IOMSD overcomes the boundary errors of image deformation,and it is invariant to changes of linear illumination and monotonic intensity. Experimental results show that IOMSD has robust performance under image rotation,viewpoint change and illumination change. Moreover,it is stabler in image deformation than MSCD. 2013 Vol. 26 (4): 409-416 [Abstract] ( 624 ) [HTML 0KB] [ PDF 1503KB] ( 810 )

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