Engineering and manufacturing applications of ART-1 neural networks

Adaptive resonance theory (ART) neural networks offer many attractive properties for applications to engineering and manufacturing problems. In this paper we discuss some of these properties and then describe some important applications. The applications discussed include retrieval of engineering designs, clustering of machines for cellular manufacturing, detection of novel events in complex data or signals, and monitoring of factory machines. The applications are described in terms of their importance, how ART-1 neural networks are utilized, and the methods used to process the input data. Techniques described include using macro-circuits of ART-1 modules to provide enhanced functionality, monitoring network activations to detect novelty, and coupling multiple ART-1 modules together to provide supervised learning.<>