Enhanced Direct Linear Discriminant Analysis for Feature Extraction on High Dimensional Data

We present an enhanced direct linear discriminant analysis (EDLDA) solution to effectively and efficiently extract discriminatory features from high dimensional data. The EDLDA integrates two types of class-wise weighting terms in estimating the average within-class and between-class scatter matrices in order to relate the resulting Fisher criterion more closely to the minimization of classification error. Furthermore, the extracted discriminant features are weighted by mutual information between features and class labels. Experimental results on four biometric datasets demonstrate the promising performance of the proposed method.