Abstract

Adaptive Resonance Theory (ART) was introduced by Steven Grossberg as a theory of human cognitive information processing (Grossberg 1976, 1980). Extending the capabilities of the ART 1 model, which can learn to categorize patterns in binary data, fuzzy ART as described in (Carpenter, Grossberg, and Rosen 1991) has become one of the most commenly used Adaptive Resonance Theory models (Brito da Silva, Elnabarawy, and Wunsch 2019). By incorporating fuzzy set theroy operators, fuzzy ART is capable of learning from binaray and bounded real valued data. Its advantage over other unsupervised learning algorithms lies in the flexibility of the learning rule. If a given input feature does not resemble a known category satisfactorily, as determined by the vigilance test, a new category is initialized. Hence, the total number of categories (or clusters) is not determined a-priori, like k-means, but chosen in accordance with the data and the context of already learnt representations. This vignette explores the use of the fuzzy ART implementation as provided by the FuzzyART R package.

Department(s)

Electrical and Computer Engineering

Research Center/Lab(s)

Applied Computational Intelligence Lab (ACIL)

Document Type

Documentation

Document Version

Preprint

File Type

text

Language(s)

English

Rights

© 2021 The Authors, All rights reserved.

Publication Date

01 May 2021

Share

 
COinS