Laplacian Maximum Scatter Difference Discriminant Criterion

Linear Laplacian Discrimination (LLD) as a non-linear feature extraction method has obtained very extensive applications. However, LDD suffers from the small sample size problem (SSS) and/or the type of the sample space when it is used. In order to circumvent such shortcomings, in this paper a Laplacian Maximum Scatter Difference Discriminant Criterion (LMSDC) is proposed by using new contextual-distance metric and integrating maximum scatter difference discriminant criterion(MSDC) into LDD. The proposed criterion can obviously decrease the dependence on the sample space and solve small sample size problem. The experimental results indicate the above advantages of the proposed method LMSDC.

[1]  John Wright,et al.  Segmentation of Multivariate Mixed Data via Lossy Data Coding and Compression , 2007, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[2]  I. Jolliffe.  Principal Component Analysis , 2002 .

[3]  Deli Zhao,et al.  Linear Laplacian Discrimination for Feature Extraction , 2007, 2007 IEEE Conference on Computer Vision and Pattern Recognition.

[4]  Catherine Blake,et al.  UCI Repository of machine learning databases , 1998 .

[5]  Deli Zhao,et al.  Laplacian PCA and Its Applications , 2007, 2007 IEEE 11th International Conference on Computer Vision.

[6]  Chia-Wei Hsu,et al.  A Linear Feature Extraction for Multiclass Classification Problems Based on Class Mean and Covariance Discriminant Information , 2006, IEEE Trans. Pattern Anal. Mach. Intell..

[7]  Xiaofei He,et al.  Locality Preserving Projections , 2003, NIPS.

[8]  Deli Zhao,et al.  Contextual Distance for Data Perception , 2007, 2007 IEEE 11th International Conference on Computer Vision.

[9]  R. Fisher.  THE USE OF MULTIPLE MEASUREMENTS IN TAXONOMIC PROBLEMS , 1936 .