Robust tool for feature extraction and its application

The aim of this paper is to present a new robust feature extraction method. Our method is an extension of the classical Partial Least Squares (PLS) algorithm. However, a robust approach and new weighted separation criterion is applied. This algorithm based on Minimum Covariance Determinant (MCD) approach and new separation criterion called Weighted Criterion of Difference Scatter Matrices (WCDSM). The new separation criterion uses the weighted difference between within and between scatter matrices to measure the separation between classes. Designed algorithm can distinguish between samples from two classes. This algorithm can be applied to low- and high dimensional data variables, and to one or multiple response variables. In order to compare the performance of the classification the economical datasets are used.