Selection Of Informative Features using The Modified Version Of The Delta Method

Abstract

Currently, the most common criteria for informative features are heuristic criteria related to assessing the separability of given classes and based on the compactness hypothesis fundamental in pattern recognition: with increasing distance between classes, their separability improves. “Good” are those signs that maximize this distance.Although heuristic criteria, although they are widely used in solving practical problems of classification, however, in theoretical terms they are little studied.