Fuzzy ARTMAP with explicit and implicit weights

Abstract

ARTMAP is one of the famous supervised learning systems. Many learning methods for ARTMAP have been proposed since it was devised as a system to solve Stability-Plasticity Dilemma. AL-SLMAP was implemented by slightly modifying FCSR which was the original learning method for fuzzy ARTMAP (FAM). Although AL-SLMAP can solve pattern recognition problems in a noisy environment more effectively than FCSR, AL-SLMAP is less suitable for region classification problems than FCSR. This means that AL-SLMAP has some problems which do not exist in FCSR. In this paper, we propose a learning method for FAM with explicit and implicit weights to overcome the problems. © 2008 Springer-Verlag Berlin Heidelberg.