Abstract：A method of neighbor class linear discriminant analysis (NCLDA) is proposed. Linear discriminant analysis (LDA) is a special case of this method. LDA finds the optimal projections by maximum between-class scatter while by minimum within-class scatter. The between-class scatter is an average over divergences among all classes. In NCLDA,between-class scatter is defined as average divergences between one class and its k nearest neighbor classes. By selecting proper numbers of neighbor class, NCLDA alleviates overlaps among classes caused by LDA. The experimental results show that the proposed NCLDA is robust and outperforms LDA.
[1] Liu Chenglin.Normalization-Cooperated Gradient Feature Extraction for Handwritten Character Recognition.IEEE Trans on Pattern Analysis and Machine Intelligence,2007,29(8): 1465-1469 [2] Fisher R A.The Statistical Utilization of Multiple Measurements.Annals of Eugenics,1938,8(4): 376-386 [3] Rao C R.The Utilization of Multiple Measurements in Problems of Biological Classification.Journal of Royal Statistical: Society B,1948,10(2): 159-203 [4] Loog M,Duin R,Haeb-Umbach R.Multiclass Linear Dimension Reduction by Weighted Pairwise Fisher Criteria.IEEE Trans on Pattern Analysis and Machine Intelligence,2001,23(7): 762-766 [5] Li Zhifeng,Lin Dahua,Tang Xiaoou.Nonparametric Discriminant Analysis for Face Recognition.IEEE Trans on Pattern Analysis and Machine Intelligence,2009,31(4): 755-761 [6] Tao Dacheng,Li Xuelong,Wu Xindong,et al.Geometric Mean for Subspace Selection.IEEE Trans on Pattern Analysis and Machine Intelligence,2009,31(2): 260-274 [7] Liu Chengjun,Wechsler H.Enhanced Fisher Linear Discriminant Models for Face Recognition // Proc of the 14th International Conference on Pattern Recognition.Brisbane,Australia,1998,II: 1368-1372 [8] Yu Hua,Yang Jie.A Direct LDA Algorithm for High-Dimensional Data with Application to Face Recognition.Pattern Recognition,2001,34(10): 2067-2070 [9] Fukunaga K.Introduction to Statistical Pattern Recognition.London,UK: Academic Press,1990 [10] Bressan M,Vitria J.Nonparametric Discriminant Analysis and Nearest Neighbor Classification.Pattern Recognition Letters,2003,24(15): 2743-2749 [11] Li Zhifeng,Lin Dahua,Tang Xiaoou.Nonparametric Subspace Analysis for Face Recognition // Proc of the International Conference on Computer Vision and Pattern Recognition.San Diego,USA,2005: 961-966 [12] Rueda L,Oommen B J,Henriquez C.Multi-Class Pairwise Linear Dimensionality Reduction Using Heteroscedastic Schemes.Pattern Recognition,2010,43(7): 2456-2465 [13] Wang Haixian,Chen Sibao,Hu Zilan,et al.Locality-Preserved Maximum Information Projection.IEEE Trans on Neural Networks,2008,19(4): 571-585 [14] Roweis S T,Saul L K.Nonlinear Dimensionality Reduction by Locally Linear Embedding.Science,2000,290(5500): 2323-2326 [15] Gao Tianfu,Liu Chenglin.High Accuracy Handwritten Chinese Character Recognition Using LDA-Based Compound Distances.Pattern Recognition,2008,41(11): 3442-3451 [16] Kimura F,Takashina K,Tsuruoka S,et al.Modified Quadratic Discriminant Function and the Application to Chinese Character Recognition.IEEE Trans on Pattern Analysis and Machine Intelligence,1987,9(1): 149-153 [17] Xu Bo,Huang Kaizhu,Liu Chenglin.Dimensionality Reduction by Minimal Distance Maximization // Proc of the 20th International Conference on Pattern Recognition.Istanbul,Turkey,2010: 569-572
2012, Vol. 25 
