ART I and pattern clustering algorithms

✍ Scribed by Barbara Moore

• Publisher: Elsevier Science
• Year: 1988
• Tongue: English
• Weight: 23 KB
• Volume: 1
• Category: Article
• ISSN: 0893-6080
• DOI: 10.1016/0893-6080(88)90155-4

✦ Synopsis

The ART I system developed by Gaff Carpenter and Stephen Grossberg is a dynamical system that implements a pattern clustering algorithm. In addition to grouping patterns~ the algorithm automatically determines the best number of groupings. The dynanfical systems implementation is interesting for many reasons, one of which is that it is seems easily realizable in parallel analog hardware. We derive the pattern clustering algorithm from the ART I architecture. We repeat the proofs of ART I convergence presented by Carpenter and Grossberg, but in the domain and ternfinology of pattern analysis. In this framework we examine other properties of the algorithm's behavior. Finally, we ask whether other pattern clustering algorithms might be implementable on an architecture similar to ART I.