ABSTRACT

A property of neural networks (e.g., ART models). As a network trains itself on a finite number of cases, the various weights should gradually stabilize.