A memetic fuzzy ARTMAP by a grammatical evolution approach

Abstract

This paper presents a memetic fuzzy ARTMAP (mFAM) model constructed using a grammatical evolution approach. mFAM performs adaptation through a global search with particle swarm optimization (PSO) as well as a local search with the FAM training algorithm. The search and adaptation processes of mFAM are governed by a set of grammatical rules. In the memetic framework, mFAM is constructed and it evolves with a combination of PSO and FAM learning in an arbitrary sequence. A benchmark study is carried out to evaluate and compare the classification performance between mFAM and other state-of-art methods. The results show the effectiveness of mFAM in providing more accurate prediction outcomes.