Review of Unsupervised Adaptive Resonance Theory

Abstract

Introduced by Stephen Grossberg in the year 1976 Adaptive Resonance Theory (ART), deals a wide variety of Artificial Neural Networks. Initially ART system was mainly for the unsupervised type learning but introduction of different ART systems have lead it deal the supervised learning as well. ART covers a huge range of neural networks for solving the problem of stability and plasticity. ART NN is appreciated for emerging mature clusters of random structures of input patterns by self-organizing capability. There are several types of architectures of ART neural network. This paper shows different type of ART architecture for unsupervised learning such as ART1, ART2, ART2A, ART3 and Fuzzy ART with parameters like choice, pre-processing and adaption rule. The motive of study is to support several researchers in ART architecture area in order for better understanding of suitable components while designing specific neural network classifier.