Theoretical characterizations of possibilities and impossibilities of Hopfield neural networks in solving combinatorial optimization problems
Satoshi Matsuda · 2002
The asymptotical stability and instability conditions of the solutions and nonsolutions are proved for corners of Hopfield neural hypercube in solving typical combinatorial optimization problems, i.e., traveling salesman problem, N-queen problem and Hitchcock problem. These conditions make the theoretical characterizations of many possibilities and impossibilities of Hopfield neural networks in solving combinatorial optimization problems.>