Artificial neural systems for geometric and non-geometric combinatorial optimization problems

Summary form only given, as follows. Novel neural network approaches to two combinatorial optimization problems, the maximal independent set problem (which is nongeometric) and the traveling salesman problem (the planar case is geometric). The primary contribution of the present work with respect to the maximal independent set problem is to present a neural network approach which suits especially well a specific instance of the problem which occurs in operations research. For the traveling salesman problem, a particularly simple approach akin to competitive learning with a conscience mechanism is used to generate good, feasible solutions quickly