Robust Classifiers Using Imprecise Probability Models and Importance of Classes

摘要: A framework for constructing robust classification models is proposed in the paper. An assumption about importance of one classes comparison with other incorporated into models. It often takes place real application, example, reliability, medical diagnostic, etc. main idea underlying to consider a set probability distributions on training examples produced by imprecise such as linear-vacuous mixture and constant odd-ratio contaminated Extreme points sets are tool classifiers. shown that algorithms computing optimal parameters reduced finite number weighted support vector machines weights determined extreme points. Experimental results synthetic data illustrate