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.7, Published 2014-07-30
Papers and Reports    Researches and Applications   
   
Papers and Reports
577 Vector Quantization Codebook Design and Speech Recognition Based on Immune Cat Swarm Optimization Algorithm
YANG Shu-Ying, LIU Xu-Peng, TAO Chong, LIU Ting-Ting
In the process of codebook design, traditional LBG algorithm is often used for vector quantization which depends on the initial codebook selection and easily falls into local optimum. A vector quantization codebook design method based on immune cat swarm optimization algorithm (ICSO) is proposed to solve the problems.The population is divided into searching group and tracking group. Clonal expansion operator is used for local search in the searching group, and the number of mutation individual is adjusted according to the fitness value. Moreover, dynamic vaccine extraction and vaccination operator are used for global search in the tracking group. The crossover and mutation between individual gene and vaccine make the result close to the optimal solution, and the descendant population is updated through the balance of concentration equilibrium operator and selection operator. Finally, the optimal codebook is obtained from the training vectors by the proposed algorithm and is inputted to the HMM model for training and recognition. The simulation results show that the proposed algorithm does not depend on the selection of initial codebook, has strong robustness and reduces the speech recognition error rate.
2014 Vol. 27 (7): 577-583 [Abstract] ( 572 ) [HTML 1KB] [ PDF 479KB] ( 820 )
584 Time-Varying Gene Regulatory Networks Construction Based on Logistic Regression
NI Xiao-Hong, SUN Ying-Fei
At present, network structures derived from most gene regulatory network reconstruction methods are static, which do not change with time.However, in the cell cycle or different growth stages of an organismat, the topology of regulatory network changes significantly, which makes it difficult to understand the spatial-temporal mechanism of gene regulation. Therefore, an algorithm for the network is proposed based on time lagged Mutual Information (TLMI) and a kernel-reweighted l1-regularized logistic regression model. Two biological scenarios, the developmental stages of Drosophila melanogaster and the response of Saccharomyces cerevisiae to benomyl poisoning, are analyzed. The experimental results show that the proposed method reflects the impact of different cell states of interaction between genes and effectively acquires the dynamic effect of gene regulatory networks changing with time.
2014 Vol. 27 (7): 584-550 [Abstract] ( 400 ) [HTML 1KB] [ PDF 626KB] ( 968 )
591 Regular Path Query on CP-nets
LIU Jing-Lei, LIAO Shi-Zhong
A formal preference knowledge representation framework, conditional preference networks (CP-nets), is introduced, and regular path query on the model is studied. Firstly, two kinds of queries, vertex query and path query,are given from the point of view of the database theory,and the expressive ability of preference database is proved to be stronger than that of relational database. Then, the induced reach ability relations of each atom expressions are obtained by constructing the syntax parse binary tree of regular expression, thereafter, the reach ability relation induced by regular expression of root vertex is obtained by means of dynamic programming. Moreover, the correctness and combined complexity of this algorithm are proved. Finally, a possible application scenario of regular path query is given, which demonstrates it can be applied in the preference planning sequence.
2014 Vol. 27 (7): 591-598 [Abstract] ( 455 ) [HTML 1KB] [ PDF 394KB] ( 848 )
599 Fuzzy Reasoning with Contradictory, Opposite and Medium Negation
ZHANG Sheng-Li
With the development of knowledge processing, the cognition and processing of negative knowledge receive widely attention. After extending the cognition about negation of fuzzy judgment sentences, one form of fuzzy judgment sentence with contradictory, opposite and medium negation is proposed. And the corresponding semantic description and logic description are presented. On this basis, 3 kinds of fuzzy reasoning rules with different negations are obtained after extending the truth-valued range model, and an extended Compositional Rules of Inference (CRI) algorithm and a demonstration are presented.
2014 Vol. 27 (7): 599-610 [Abstract] ( 376 ) [HTML 1KB] [ PDF 474KB] ( 781 )
Researches and Applications
611 Automatic Marker Registration and Prediction Algorithm for Optical Motion Capture System
TAN Guang-Hua, ZHOU Mei-Lan, GAO Chun-Ming, LI Ren-Fa
An automatic marker registration method is proposed to solve the registration problem of scattered point-set in optical motion capture system. Firstly, the markers in the first frame are segmented to provide the initial skeleton template with the knowledge of human body topology structure. Then, according to the local rigidity property and spatio-temporal information,the registration of marker is achieved by finding the most matching point in previous frame, which does not need specifying the initial template by hand. Under the assumption that markers do not change a lot in two consecutive frames, the coordinates of missing markers are predicted by using local rigidity property of the same segment. The experimental results show that every marker can be registered quickly and accurately and the proposed method has strong robustness and real-time performance.
2014 Vol. 27 (7): 611-616 [Abstract] ( 465 ) [HTML 1KB] [ PDF 873KB] ( 1141 )
617 SAR Target Configuration Recognition Based on Sparse Representation
LIU Ming, WU Yan, WANG Fan, ZHANG Qiang, LI Ming
Applying sparse representation (SR) to a sample, the sparse vector can be obtained, which not only sufficiently describes the characteristic of the sample, but also has the capability of discriminant. A SAR target configuration recognition algorithm based on SR is proposed. Firstly, Feature extraction is implemented to eliminate the influence of the speckle noise. Then, a dictionary is constructed by all the training samples, and the testing sample is projected onto the dictionary to obtain the sparse vector. Finally, according to the fact that the difference between the samples with the same label and closest azimuth angles is the smallest, the principle of the minimum single sample reconstruction error is established to realize SAR target configuration recognition. The experiments on the MSTAR datasets validate the effectiveness of the proposed algorithm.
2014 Vol. 27 (7): 617-622 [Abstract] ( 428 ) [HTML 1KB] [ PDF 541KB] ( 781 )
623 Ranking Topic Models without Query
XIAO Zhi-Bo, CHE Feng, WU Di, LI Qing-Feng, LU Ming-Yu
Topic models have become important tools in machine learning and natural language processing, which can discover hidden topics in large-scale corpus. However, as the size of the corpus grows, the scale of discovered topics grows. Most topic models are on the basis of bag-of-words model, and the orders between terms cannot be described, which makes topics undistinguishable from each other. Ranking topic models without query framework is proposed in this paper, in which topics are ranked to get ordered topic list according to their relationships. Topic relationships are used to evaluate topic influence in topic level, and term significance is used to evaluate term importance in term level and popular ranking topics with little semantics are weakened. Since there is no acknowledged evaluation criterion in ranking topic model, ranked topics are used as features to perform automatic summarization of multi-document, and the performance of ranking topic models are indirectly measured by summarization performance. The experimental results show that ranking topic models outperform topic models without ranking.
2014 Vol. 27 (7): 623-630 [Abstract] ( 405 ) [HTML 1KB] [ PDF 474KB] ( 933 )
631 A URL Filtering Generation Algorithm Based on Similarity Degree for Web Crawling
CHEN Hui-Hui, SHU Yun-Xing, LIN Li
Web text is an important component of the corpus, however, unnecessary time consumption for visiting redundant URLs influences the quality and efficiency of the large scale web crawling. The quality and efficiency of Web crawling can be promoted by using high effective URL filtering rules. The distribution of files in the virtual directories of a website is uneven and a URL filtering rule generation method is introduced to discover the clustering region of target files. Firstly, URLs are transformed into regular expressions and they are divided into many groups by clustering same regular expressions. Then, the similarity degrees between URLs in one group are calculated and the virtual path tree is constructed by using URLs with higher similarity degrees. Finally, the virtual path tree is utilized to generate URL filtering rules and classification rules for Web crawling. The algorithms for generating virtual path tree are introduced in detail and the experimental results of the generated virtual path trees and the filtered URLs are compared by using different similarity degree thresholds.
2014 Vol. 27 (7): 631-637 [Abstract] ( 602 ) [HTML 1KB] [ PDF 717KB] ( 14629 )
638 Privacy Preservation for the Incremental Updating Database
CHEN Wen
The research of sensitive pattern hiding is important in privacy preserving data mining. Existing sensitive pattern hiding algorithms are originally designed for static database which cannot handle incremental datasets effectively and efficiently. To hide sensitive patterns in the incremental environment, a selection strategy for optimal victim items with minimal edge effect is designed based on sensitive pattern graph and a privacy preservation algorithm in the incremental updating database is proposed. The instance analysis and experimental results validate the correctness, efficiency and scalability of the proposed method.
2014 Vol. 27 (7): 638-645 [Abstract] ( 424 ) [HTML 1KB] [ PDF 567KB] ( 711 )
646 A Semi-Supervised Text Clustering Based on Strong Classification Features Affinity Propagation
WEN Han, XIAO Nan-Feng
A semi-supervised text clustering based on strong classification features affinity propagation (SCFAP) is proposed to handle spare document data with large scale and high dimensions. In the clustering process, strong classification features are extracted to construct a reasonable similarity measure by using a small amount of labeled samples. Moreover, in order to improve the execution efficiency of the algorithm, the unlabeled documents with maximum category certainty are transferred from unlabeled collection to labeled collection in each round of iteration. The experimental results show that the improvement is greatly helpful to upgrade the performance and accuracy of the classical affinity propagation algorithm. The SCFAP algorithm shows better applicability on Reuter-21578 and 20 Newsgroups. The micro average Fμ index and the clustering purity index are synthetically observed, the semi-supervised text clustering algorithm based on SCFAP can get better clustering results rapidly.
2014 Vol. 27 (7): 646-654 [Abstract] ( 384 ) [HTML 1KB] [ PDF 639KB] ( 768 )
655 Road Detection Method Based on Graph Model
BAI Meng, LI Min-Hua
To solve the detection problem for low-speed mobile robots in urban environments, a road detection approach based on graph model is proposed to detect road region. Firstly, the road image is segmented into sub-images,and the feature vector of each sub-image is computed to generate node set of the graph model. Then, the concept of neighbor radius for nodes is proposed to calculate edge weights between adjacent nodes, from which the edge set of graph is acquired. The graph nodes merging rule based on the minimum spanning tree algorithm is used to combine nodes, which realizes the road image segmentation. Finally, the road nodes are extracted to segment the road region by setting a node sample window. The relationship among the road detection precision, sub-image size and threshold parameter is studied in the experiment. The feasibility of using gray feature to detect road region is verified. The experimental results demonstrate that the proposed approach can detect road regions from different kinds of urban road images effectively and is suitable for the road detection.
2014 Vol. 27 (7): 655-662 [Abstract] ( 508 ) [HTML 1KB] [ PDF 2224KB] ( 812 )
663 A Semi-Structured Tibetan Text Clustering Algorithm Based on Swarm Intelligence
KANG Jian, QIAO Shao-Jie, GESANG Duoji, HAN Nan, HONG Xi-Jin, NIMA Zhaxi, FAN Xiao-Gang

To apply swarm intelligence techniques to cluster semi-structured Tibetan Web texts, a semi-structured Tibetan text clustering algorithm based on swarm Intelligence (SCAST) is proposed. Taking into a full consideration of accuracy and efficiency of Tibetan text clustering, a vector space model is used to express Tibetan texts, and the Tibetan texts and intelligent ants are randomly put in a two dimensional text vector space. Then, intelligent ants randomly select a Tibetan text, calculate the similarity between this text and others in the local area,and compute the probability of pick-up operation or drop-down operation to determine whether to pick up, move, or drop down the text. Finally, Tibetan texts are accurately clustered according to their similarities by iterative training of the proposed algorithm. The experimental results on real Tibetan Web text datasets show that the proposed algorithm is more accurate than the traditional k-means clustering algorithm with average increase of 8.0%.
2014 Vol. 27 (7): 663-672 [Abstract] ( 488 ) [HTML 1KB] [ PDF 1001KB] ( 778 )
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