Mapping combinatorial optimization problems onto neural networks

Neural networks have been proposed as a model of computation for solving a wide variety of problems in elds as diverse as combinatorial optimization, vision, and pattern recognition. The ability to map and solve a number of interesting problems on neural networks motivates a proposal for using neural networks as a highly parallel model for general-purpose computing. We review this proposal , and show how to map a number of interesting combinatorial optimization problems from graph theory, VLSI placement, and operations research.