Using a Genetic Algorithm to tune Potts Neural Networks

This paper describes the successful application of a Genetic Algorithm to tuning the parameters of a Potts Neural Network.Potts Neural Networks can be used to solve optimization problems. First the problem is mapped upon the network by generating neurons and connections depending on the instance of the problem at hand. The stable state found with simulation represents a solution for the problem.There are several parameters that influence correctness and quality of the solutions found. The parameters depend on the problem to be solved. Another optimization technique inspired by nature, Genetic Algorithms, is used to find parameter settings that give the best performance of the network.The approach has been tested on two applications of Potts Neural Networks. In both cases good results were achieved.