Computing with structured neural networks

Abstract

The authors are concerned with how one can design, realize, and analyze networks that embody the specific computational structures needed to solve hard problems. They focus on the design and use of massively parallel connectionist computational models, particularly in artificial intelligence. They describe a computing environment for working with structured networks and present some sample applications. Throughout, they treat adaptation and learning as ways to improve structured networks, not as replacements for analysis and design.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>