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....

2015 Vol.28 Issue.10, Published 2015-10-30 Papers and Reports    Researches and Application        Papers and Reports 865 Cost-Driven Scheduling Strategy for Scientific Workflow under Multi-cloud Environment LIN Bing, GUO Wen-Zhong, CHEN Guo-Long, CHEN Huang-Ning Aiming at the deadline-constrained scientific workflow scheduling problem under multi-cloud environment, the concept of partial critical paths algorithm is introduced. A cost-driven scheduling strategy for scientific workflow is proposed to reduce the execution cost of workflow as much as possible and satisfy its deadline constraint. The characteristics of multi-cloud environment and scientific workflows are taken into account in this strategy. Firstly, the adjacent two tasks with a common directed cut-edge are merged into a single task based on the workflow structure. Then, the partial critical paths with subdeadline constraints are searched based on the critical parent iterative mechanism. Finally, the most suitable instances are allocated to the partial critical path and all the tasks in the path are scheduled to their corresponding instance. Various workflows are used for evaluating the proposed strategy and experimental results show that the proposed strategy has a better execution efficiency and a lower workflow execution cost. 2015 Vol. 28 (10): 865-875 [Abstract] ( 540 ) [HTML 1KB] [ PDF 807KB] ( 939 ) 876 Sparse Stochastic Optimization Algorithm with Optimal Individual Convergence Rate Based on Random Step-Size ZHOU Bai, TAO Qing, CHU De-Jun Almost all sparse stochastic algorithms are developed from the online setting, and only the convergence rate of average output can be obtained. The optimal rate for strongly convex optimization problems can not be reached as well. The stochastic optimization algorithms are directly studied instead of the online to batch conversation in this paper. Firstly, by incorporating the L2 regularizer into the L1-regularized sparse optimization problems, the strong convexity can be obtained. Then, by introducing the random step-size strategy from the black-box optimization method to the state-of-the-art algorithm-composite objective mirror descent (COMID), a sparse stochastic optimization algorithm based introducing on random step-size hybrid regularized mirror descent (RS-HMD) is achieved. Finally, based on the analysis of characteristics of soft threshold methods in solving the L1-regularized problem, the optimal individual convergence rate is proved. Experimental results demonstrate that sparsity of RS-HMD is better than that of COMID. 2015 Vol. 28 (10): 876-885 [Abstract] ( 420 ) [HTML 1KB] [ PDF 597KB] ( 537 ) 886 Vertical Evolution of Decision Evolution Sets HU YU-Wen, XU Jiu-Cheng, ZHANG Qian-Qian The study on decision evolution sets is limited to horizontal evolution and there is a lack of the study on vertical evolution. In this paper, decision information system is divided by different granularities based on decision evolution sets, and therefore different granule sets from the decision information system are formed in different granular spaces. Then, the interaction and the relationship of the decision rules generated from the time granules in different granular spaces are analyzed. Finally, the running process of vertical evolution of decision information system is shown by examples. 2015 Vol. 28 (10): 886-895 [Abstract] ( 422 ) [HTML 1KB] [ PDF 396KB] ( 424 ) 896 Linear Fix-Point Neighbor Embedding Analysis Method QIU Hong, WANG Wan-Liang, ZHENG Jian-Wei To solve the out-of-sample problem of t-distributed stochastic neighbor embedding(t-SNE) analysis method and overcome the unfeasibility of manually adjusting the involved parameters in practice, a linear fix-point neighbor embedding(LFNE) analysis method is proposed based on a fix-point optimization algorithm. Based on t-SNE, the linear projection matrix is introduced to reveal the underlying structure of data manifold in LFNE. Then, the penalty function is built by minimizing the Kullback-Leibler divergence of original space and subspace. Furthermore, the efficiency and the robustness of LFNE optimization are improved by the fix-point optimization algorithm. The proposed method is evaluated on artificial synthetic data and COIL-20 database. Experimental results demonstrate the better effectiveness of visualization by LFNF. 2015 Vol. 28 (10): 896-902 [Abstract] ( 397 ) [HTML 1KB] [ PDF 1907KB] ( 708 ) 903 Soft Subspace Clustering Based on Particle Swarm Optimization QIU Yun-Fei, YANG Qian, TANG Xiao-Liang The performance of soft subspace clustering depends on the objective function and subspace search strategy, and cluster validity analysis is the main indicator of its performance. Aiming at the subspace clustering performance, a soft subspace clustering algorithm based on particle swarm optimization (SC-PSO) is proposed. Firstly, combining inter-cluster separation with feature weight based on K means-type clustering framework, a fuzzy weighting soft subspace objective function is designed. Then, particle swarm optimization with inertia weight is used as a subspace search strategy to jump out of the local optimum. Finally, the optimal cluster number is selected by the proposed cluster validity function.The experimental results demonstrate that SC-PSO improves the clustering accuracy and automatically determines the optimal cluster number. 2015 Vol. 28 (10): 903-912 [Abstract] ( 540 ) [HTML 1KB] [ PDF 579KB] ( 637 ) 913 Recommendation Algorithm Based on User-Interest-Item Tripartite Graph ZHANG Yan-Mei, WANG Lu, CAO Huai-Hu, MAO Guo-Jun Since most of the existing personalized recommendation algorithms pursue a higher accuracy, their performance is affected by the problem of user interest over-specialization. An algorithm is proposed to fully mine and use the implicit user interest information for recommendation. The probabilistic topic model is adopted to extract user interest distribution, and the weighted tripartite graph of user-interest-item is generated. Then the user item resource value is allocated by material diffusion algorithm in user-interest and interest-item bipartite graphs respectively, and the Top-K recommendation list is generated according to the rank of item resource values. Experimental results on Movielens datasets show the proposed algorithm relieves the problem of user interest over-specialization. Meanwhile the recommendation accuracy is improved . 2015 Vol. 28 (10): 913-921 [Abstract] ( 629 ) [HTML 1KB] [ PDF 542KB] ( 843 ) Researches and Application 922 Improved Associative Classification Algorithm for Multiclass Imbalanced Datasets HUANG Zai-Xiang, ZHOU Zhong-Mei, HE Tian-Zhong, Zheng Yi-Feng Instances in some classes are rare in multiclass imbalanced datasets and therefore few rules for these classes are generated by support-confidence based associative classification algorithms. Consequently, instances in these minority classes are difficult to be correctly classified. Aiming at this problem, an improved associative classification algorithm for multiclass imbalanced datasets is proposed. To extract more rules for minority classes, rules are extracted according to positive correlation between itemsets and classes. Then, to improve the priority of minority classes rules, the rule strength based on itemsets class distribution is designed to rank rules. Finally, to address problems of no matched rules or matched rules in conflict, a k nearest neighbor algorithm is incorporated into the improved associative classification to classify new instances. Experimental results show that the proposed algorithm extracts more minority classes rules and promotes the priority of the minority classes rules compared with support-confidence based associative classification, and thus G-mean and F-score value for multiclass imbalance datasets are improved. 2015 Vol. 28 (10): 922-929 [Abstract] ( 437 ) [HTML 1KB] [ PDF 438KB] ( 573 ) 930 Dynamic Adaptive Ant Colony Optimization Algorithm for Min-Max Vehicle Routing Problem GE Bin, HAN Jiang-Hong, WEI Zhen, CHENG Lei, HAN Yue To solve the min-max vehicle routing problem (MMVRP), a dynamic adaptive ant colony optimization algorithm is proposed. The dynamic max-min ant system is adopted to adjust the optimal solution. τmin is updated per iteration, it is regarded as the function of maximum in the pheromone matrix, and the probability of selecting arc is adjusted according to the optimal arc. A kind of gray model is employed to forecast and control the boundary of pheromone matrix to enhance the self-adaption of parameters in ant colony algorithm. Advantage of pheromone associated with accumulation rules is taken to update multiple nodes with relatively high concentration of pheromone and edges nearby. The proposed algorithm is tested on examples. The simulation results show that compared with linear programming algorithm and other related ant colony algorithms, the proposed algorithm has a higher convergence speed and better optimization performance and applicability. 2015 Vol. 28 (10): 930-938 [Abstract] ( 526 ) [HTML 1KB] [ PDF 550KB] ( 908 ) 939 Human Action Description Algorithm Based on Depth Dense Spatio-Temporal Interest Points SONG Jian-Ming, ZHANG Hua, GAO Zan, ZHANG Yan, XUE Yan-Bing, XU Guang-Ping Much attention is paid to action description algorithm based on depth data now. However, there is no robust, efficient and distinguishing feature representation for depth data. To solve the problem, human action description algorithm based on depth dense spatio-temporal interest point is proposed. Multi-scale depth dense feature spatio-temporal interest points are selected and then tracked, and the trajectories of these points are saved. Finally, the trajectory information is utilized to represent human action. Through the evaluation on DHA, MSR Action 3D and UTKinect depth action dataset, the proposed algorithm show better performance compared with some state-of-the-art algorithms. 2015 Vol. 28 (10): 939-945 [Abstract] ( 406 ) [HTML 1KB] [ PDF 591KB] ( 858 ) 946 Image Recognition Algorithm Based on Log-Gabor Wavelet and Riemannian Manifold Learning LIU Yuan, WU Xiao-Jun In image recognition applications, Riemannian manifold learning algorithms can not eliminate the redundant information in images effectively. Therefore, an image recognition algorithm based on Log-Gabor wavelet and Riemannian manifold learning is presented. Firstly, images are processed by the Log-Gabor filter to obtain high-dimensional Log-Gabor image features. Then, the Riemannian manifold learning algorithm is used to reduce the dimensionality of the image features. Research shows that the integration of Log-Gabor wavelet and Riemannian manifold learning is in accord with the process of human visual perception. The proposed algorithm has better robustness to illumination and angle variation of the image. Experimental results on several standard databases indicate the effectiveness of the proposedalgorithm. 2015 Vol. 28 (10): 946-952 [Abstract] ( 575 ) [HTML 1KB] [ PDF 544KB] ( 847 ) 953 Stochastic Particle Swarm Optimization Algorithm with Embedded Cascading Chaotic Strategy LI Sheng, HE Ming-Hui, LI Jian-Lin, ZHANG Li Aiming at the premature convergence problem frequently occurring in complex optimization problems of particle swarm optimization (PSO) algorithms, an iteration equation of stochastic particle swarm evolution is proposed by importing the disturbed track factor. The statistical behavior of particles produced by the equation is ensured to approach certain convergence centre, while their dependence on the address of previous generation appears to be stochastic.Thus, it is possible for particle swarm to jump quickly or migrate instead of being trapped in the local extremum at early evolution. Meanwhile, the cascading chaotic strategy and the symmetric extremum perturbation strategy are employed to further enhance local convergence velocity and globe search capacities, respectively. Experimental results indicate that stochastic chaotic PSO algorithm composed by the proposed equation and the improved strategies is better than other homologous particle swarm optimization algorithms in robustness, convergence speed and accuracy. 2015 Vol. 28 (10): 953-960 [Abstract] ( 453 ) [HTML 1KB] [ PDF 576KB] ( 565 )

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