An Algorithm of Classification for Nonstationary Case

Piotr Kulczycki, Piotr Andrzej Kowalski

Abstract: The paper deals with the classification task, where patterns are nonstationary. The method ensures the minimum expected value of misclassifications and is independent of patterns’ shapes. This procedure eliminates elements of patterns with insignificant or even negative influence on the results’ accuracy. Appropriate modifications follow the classifier parameters, which increases the effectiveness of procedure adaptation for nonstationary patterns. The number of patterns is not methodologically limited in the presented concept.