Improvement of Fuzzy ARTMAP Performance in Noisy Input Environment Using Weighted-Average Learning

Jae Sul LEE  Chang Joo LEE  Choong Woong LEE  

IEICE TRANSACTIONS on Fundamentals of Electronics, Communications and Computer Sciences   Vol.E80-A   No.5   pp.932-935
Publication Date: 1997/05/25
Online ISSN: 
Print ISSN: 0916-8508
Type of Manuscript: LETTER
Category: Neural Networks
fuzzy ARTMAP,  weighted average,  pattern recognition,  weight vector,  category proliferation,  

Full Text: PDF>>
Buy this Article

An effective learning method for the fuzzy ARTMAP in the recognition of noisy input patterns is presented. the weight vectors of the system are updated using the weighted average of the noisy input vector and the weight vector itself. This method leads to stable learning and prevents the excessive update of the weight vectors which may cause performance degradation. Simulation results show that the proposed method not only reduces the generation of spurious categories, but aloso increases the recognition ratio in the noisy environment.