NIRS: Large scale ART-1 neural architectures for engineering design retrieval

Abstract We describe a neural information retrieval system developed for retrieval of engineering designs. Two-dimensional (2-D) and three-dimensional (3-D) representations of engineering designs are input to adaptive resonance theory (ART-1) neural networks to produce groups or clusters of similar parts. ART-1 networks are first trained to cluster designs into families, and then to recall a family of similar parts when queried with a new part design. This application is of great practical value to industry because it aids in the identification, retrieval, and reuse of engineering designs, potentially saving large amounts of nonrecurring costs. In this paper, we describe the application, the neural architectures and algorithms, the current status, and the lessons learned in developing a neural network system for production use in industry.