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....

2017 Vol.30 Issue.4, Published 2017-04-30 Papers and Reports    Researches and Applications        Papers and Reports 289 Truth Theory of Proposition Logic under Random Fuzzy Environment WU Xia, ZHANG Jialu The concept of random fuzzy truth degree of logic formulas is proposed by virtue of probability distribution on real unit interval [0,1] . It is pointed out that the random fuzzy is the common spread of truths in the valuation domain of logic formulas. Then, the concept of random fuzzy similarity degree between two logic formulas is proposed from the concept of random fuzzy truth degree. Based on it, the pseudo-metric named random fuzzy pseudo-metric is introduced on all formula sets. And it is proved that there are not isolated points in the random fuzzy logic pseudo-metric space. Moreover, by using of the integral convergence theorem in probability theory, a limit theorem of random truth degree is proved. The connection of truth degrees is illustrated by this limit theorem. Furthermore, the continuity of the logical operation in the random logic pseudo-metric space is certified and the fundamental theorems of probabilistic logic are expanded to multi-valued propositional logic. Finally, two kinds of approximate reasoning models are presented and applied to approximate reasoning of the practical problems in random logic pseudo-metric space. 2017 Vol. 30 (4): 289-301 [Abstract] ( 530 ) [HTML 1KB] [ PDF 623KB] ( 470 ) 302 Visual Tracking via Hierarchical Extreme Learning Machine and Local Sparse Model SUN Rui, ZHANG Dongdong, GAO Jun To address problems of appearance change and partial occlusion in target tracking, a tracking algorithm is presented via combing hierarchical extreme learning machine(HELM) and adaptive structural local sparse appearance model(ASLSAM). HELM is capable of extracting robust features and fast classification. ASLSAM can improve the tracking accuracy and handle the partial occlusion. Finally, results of both qualitative and quantitative evaluations on challenging benchmark image sequences demonstrate that the tracking process of the proposed algorithm is stable with high tacking precision. 2017 Vol. 30 (4): 302-313 [Abstract] ( 640 ) [HTML 1KB] [ PDF 2957KB] ( 491 ) 314 Multivariate Time Series Similarity Matching Method Based on Weighted Dynamic Time Warping Algorithm YE Yanqing, YANG Kewei, JIANG Jiang, GE Bingfeng, DOU Yajie In most of the current methods, the close correlation between variables and the shape characteristics of time series is neglected. In this paper, a similarity matching method for multivariate time series is proposed based on combined principal component analysis method and a shape-based improved weighted dynamic time warping algorithm(CPCA-SWDTW). Firstly, a shape coefficient is introduced and a shape based weighted dynamic time warping(SWDTW) algorithm is presented. Next, the principal components of multivariate time series are extracted as the representation, and thus the variable correlations can be eliminated. Besides, the variance devoting rate of each principal component is considered as the weight of each series. On the basis of the proposed representation, SWDTW is used to measure the similarity between transformed multivariate time series. Finally, the results of similarity search experiment show that CPCA-SWDTW is more efficient and robust. Moreover, the parameter sensitivity analysis experiment show that CPCA-SWDTW can be affected by the parameters in weight function to some extent. 2017 Vol. 30 (4): 314-327 [Abstract] ( 928 ) [HTML 1KB] [ PDF 4802KB] ( 605 ) 328 Local Discriminant Projection via Random Subspace HAN Lu, WU Fei, JING Xiaoyuan High dimensional data is sensitive to noise and the curse of dimensionality problem appears easily. A local discriminant projection algorithm based on random subspace(RSLDP) is proposed. The attributes of original high dimensional data are selected by random subspace method to generate a low dimensional subspace, and the nearest neighbor graphs are constructed in the low dimensional subspace. Thus, the influence of noise is reduced effectively. By RSLDP, the local inter-class weighted scatter is maximized, the local intra-class weighted scatter is minimized, and simultaneously the local scatter on data is minimized as well. Consequently, the performance of local maximal margin discriminant embedding (LMMDE) algorithm is improved.The relationship between the focusing point and its inter-class/intra-class nearest neighbor center point is well characterized by RSLDP. The effectiveness of the proposed algorithm is verified by the experiments on CMU PIE and AR face datasets. 2017 Vol. 30 (4): 328-334 [Abstract] ( 525 ) [HTML 1KB] [ PDF 595KB] ( 684 ) 335 Hybrid Recommender Algorithm Based on Graph ZHANG Yihao, LIU Xiaoyang, LIU Wanping, ZHU Changpeng Hybrid recommender is a significant way for solving the defect of various single recommender methods.A hybrid recommender algorithm based on graph is proposed in this paper.Various recommended factors are fused into graph to produce the final recommendation results. The similarity between items is calculated using the content attribute of recommended items to build correlation matrix of the nearest graph. The item profile is constructed according to the scored record of item to generate a vector function. Grounded on the above, a regular framework is used to build a graph-based learning model by combining correlation matrix and vector function and realize a personalization recommendation based on graph. By the experiments on MovieLens datasets and transaction data of Amazon online mart, the effectiveness of the proposed algorithm is verified. 2017 Vol. 30 (4): 335-342 [Abstract] ( 694 ) [HTML 1KB] [ PDF 738KB] ( 1947 ) Researches and Applications 343 One Sample per Person Face Recognition Based on Deep Autoencoder ZHANG Yan, PENG Hua Since there is only one sample for each subject, it is hard to describe intra-class variations of the subject. The performance of state-of-the-art face recognition algorithms declines in one sample per person(OSPP) face recognition. In this paper, an OSPP face recognition algorithm based on deep autoencoder(OSPP-DA) is proposed. In OSPP-DA, deep autoencoder is trained by all the images in the gallery firstly, and a generalized deep autoencoder(GDA) is generated. Then, the GDA is fine-tuned by the single sample of the subject, and a class-specified deep autoencoder(CDA) is obtained. For classification, query images are input to CDAs and the reconstruction samples of the corresponding subjects have the same intra-class variation as query images. A Softmax regression model is trained by the reconstruction samples and the query images are identified by the Softmax regression model. Experiments on public testing database are conducted and the results show the validity of OSPP-DA. Compared with some state-of-the-art algorithms, the proposed algorithm produces better performance with less time. 2017 Vol. 30 (4): 343-352 [Abstract] ( 884 ) [HTML 1KB] [ PDF 962KB] ( 638 ) 353 Prenatal Diagnosis Method of Placenta Accreta Based on Hidden Markov Model ZHANG Dong, CHEN Kai, YAN Jianying, ZHU Danhong, YE Dongyi Placenta accreta is one of the most serious complications of obstetrics. As a gold standard, the postnatal pathological examination has hysteresis and limitation. In this paper, the multi-feature associations of medical history information and color Doppler ultrasound data are used as observation sequences and the postpartum pathological results are used as hidden state sequences. The prenatal prediction method of placenta accreta based on hidden Markov model is proposed. The algorithm of Gini is used to extract the disease factors. Then, the hidden Markov model is built by the set of factors. Through the observation and hidden sequences, the prenatal prediction of placenta accreta is accomplished using Baum-Welch and Viterbi algorithms. The experimental results show that the proposed method achieves better diagnostic accuracy, sensitivity and specificity. 2017 Vol. 30 (4): 353-358 [Abstract] ( 517 ) [HTML 1KB] [ PDF 719KB] ( 414 ) 359 Towards End to End Speech Recognition System for Tibetan WANG Qingnan, GUO Wu, XIE Chuandong End to end speech recognition based on connectionist temporal classification (CTC) is applied to the Tibetan automatic speech recognition(ASR), and the performance is better than that of the state-of-the-art bidirectional long short-term memory approach. In end to end speech recognition,the linguistic knowledge such as pronunciation lexicon is not essential, and therefore the performance of the ASR systems based on CTC is weaker than that of the baseline. Aiming at this problem, a strategy combining the existing linguistic knowledge and the acoustic modeling based on CTC is proposed, and the tri-phone is taken as the basic units in acoustic modeling. Thus, the sparse problem of the modeling unit is effectively solved, and the discrimination and robustness of the CTC model are improved substantially.Results on the test set of Tibetan corpus show that the word accuracy of the model based on CTC is improved substantially and the effectiveness of the combination of the linguistic information and the CTC modeling is verified. 2017 Vol. 30 (4): 359-364 [Abstract] ( 894 ) [HTML 1KB] [ PDF 555KB] ( 1108 ) 365 Association Rules Mining Based on Multi-objective Fireworks Optimization Algorithm WU Qiong, ZENG Qingpeng According to characteristics of quantitative association rules, a quantitative association rules mining algorithm based on multi-objective fireworks optimization algorithm and opposition-based learning(QAR_MOFWA_OBL) is proposed. Firstly, fireworks optimization algorithm is utilized for a complete search of association rules. Next, opposition-based learning(OBL) is introduced to improve the convergence speed of the algorithm and reduce the probability of falling into local optimum. Then, the diversity of rules is maintained by means of the elimination mechanism of redundancy. Finally, after several iterations, the association rule set is obtained. Moreover, the thresholds of support or confidence of the proposed algorithm are not expected to be specified artificially. Simulation experiment shows the stable results are obtained on different real-world datasets, and the dataset can be adequately covered with a good balance among reliability, relevance and comprehensibility. 2017 Vol. 30 (4): 365-376 [Abstract] ( 549 ) [HTML 1KB] [ PDF 845KB] ( 493 ) 377 Riemannian Manifold Image Set Classification Algorithm Based on Log-Gabor Wavelet Features WANG Rui, WU Xiaojun The perception theory of biological neurology coincides with Riemannian manifold, and Log-Gabor filter is more suitable for nonlinear human eye logarithmic characteristic than other filters.Therefore,the combination of Log-Gabor wavelet and Riemannian manifold accords with the process of human visual perception. Grounded on covariance discriminative learning(CDL), the Riemannian manifold image set classification algorithm based on Log-Gabor Wavelet features is presented.Each image is processed by Log-Gabor filter to get its multi-scale and multi-direction features. The two-directional two-dimensional principal component analysis is adopted to reduce the dimension of covariance matrix and then the covariance discriminative learning algorithm is applied for classification.The experimental results of the proposed algorithm on several standard datasets show the superiority of the algorithm in accuracy over state-of-the-art algorithms. 2017 Vol. 30 (4): 377-384 [Abstract] ( 778 ) [HTML 1KB] [ PDF 619KB] ( 819 )

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