Entity

Time filter

Source Type


Chen S.-B.,Anhui Science and Technology University | Chen S.-B.,Key Laboratory for Industrial Image Processing and Analysis of Anhui Province | Zhao L.,Anhui Science and Technology University | Zhao L.,Key Laboratory for Industrial Image Processing and Analysis of Anhui Province | And 2 more authors.
Huanan Ligong Daxue Xuebao/Journal of South China University of Technology (Natural Science) | Year: 2014

The selection of dictionary is crucial to sparse representation classification. In order to preserve the local information of original training samples with less dictionary atoms and include more discriminant information in the learned dictionary, a new dictionary learning method based on the locality preserving criterion is proposed for sparse representation. In this method, the locality preserving criterion is imposed on coding coefficients, which makes the coding coefficients of neighboring data points in the dictionary close to each other and preserves the local information of original training samples. Experimental results on extended YaleB, AR and COIL20 databases show that the proposed method is effective because it is of higher classification performance than other methods. Source


Chen S.-B.,Anhui Science and Technology University | Chen S.-B.,Key Laboratory for Industrial Image Processing and Analysis of Anhui Province | Chen D.-R.,Anhui Science and Technology University | Chen D.-R.,Key Laboratory for Industrial Image Processing and Analysis of Anhui Province | And 2 more authors.
Tien Tzu Hsueh Pao/Acta Electronica Sinica | Year: 2016

When performing dimensionality reduction with linear projections,maximum margin criterion (MMC) is often affected by outliers and noises due to L2-norm.In this paper,L1-norm-based maximum margin criterion (MMC-L1) is proposed for dimensionality reduction.It makes full use of Maximum Margin Criterion and strong robustness of L1-norm to outliers and noises.A rapid iterative optimization algorithm,with its proof of monotonic convergence to local optimum,is given.Experiments on several public image databases verify the robustness and efficiency of the proposed method. © 2016, Chinese Institute of Electronics. All right reserved. Source


Chen S.-B.,Anhui Science and Technology University | Chen S.-B.,Key Laboratory for Industrial Image Processing and Analysis of Anhui Province | Zhao L.,Anhui Science and Technology University | Zhao L.,Key Laboratory for Industrial Image Processing and Analysis of Anhui Province | And 2 more authors.
Zidonghua Xuebao/Acta Automatica Sinica | Year: 2014

In order to further improve the classification performance via kernel tricks, two new kernel dictionary learning methods are proposed for sparse representation, which are extended from dictionary learning via locality preserving for sparse representation (LPDL). First, the original training data are projected into a high dimensional kernel space, then locality preserving based kernel dictionary learning for sparse representation (LPKDL) is proposed. Second, the kernelized locality preserving criterion is imposed on the sparse coefficients, and then the kernelized locality preserving based kernel dictionary learning for sparse representation (KLPKDL) is proposed. Experimental results show that the proposed methods are superior to other methods on classification performances. Copyright © 2014 Acta Automatica Sinica. All rights reserved. Source


Chen S.-B.,Anhui Science and Technology University | Chen S.-B.,Key Laboratory for Industrial Image Processing and Analysis of Anhui Province | Xu L.-X.,Anhui Science and Technology University | Xu L.-X.,Key Laboratory for Industrial Image Processing and Analysis of Anhui Province | And 2 more authors.
Tien Tzu Hsueh Pao/Acta Electronica Sinica | Year: 2014

Sparse representation based classification (SRC) and kernel methods are applied in many pattern recognition problems. In order to improve the classification accuracy, we propose multiple kernel sparse representation based classification (MKSRC). A fast optimization iteration method to solve sparse coefficients and the associated convergence proof to global optimal solution are given. In order to update the kernel weights of MKSRC, two different updating methods and the associated comparison are given. The experimental results on three face image databases show the superiority of the proposed multiple kernel sparse representation based classification. ©, 2014, Chinese Institute of Electronics. All right reserved. Source


Tang J.,Anhui Science and Technology University | Tang J.,Key Laboratory for Industrial Image Processing and Analysis of Anhui Province | Huang L.-L.,Anhui Science and Technology University | Huang L.-L.,Key Laboratory for Industrial Image Processing and Analysis of Anhui Province | And 4 more authors.
Huanan Ligong Daxue Xuebao/Journal of South China University of Technology (Natural Science) | Year: 2012

Aiming at the co-occurrence and relevance among the multi-label data, a novel multi-label classification algorithm using adaptive linear regression is proposed. In the algorithm, first, the classical linear regression theory is extended to the multi-label linear regression. Then, the threshold for the regression results is set by combining various evaluation criteria, thus adaptively predicting the final labels. The proposed algorithm considers not only the fixed threshold corresponding to the averages but also the adaptive thresholds reflecting the comprehensive effects of the classifier, thus reducing the influence of the distribution and noise of original data on the classification. Experimental results demonstrate that the proposed algorithm is effective in the multi-label classification. Source

Discover hidden collaborations