Integrated Genetic Algorithm and Artificial Neural Network

Genetic algorithm is used to randomly search the feasible region in global optimization problems. The search mechanism of genetic algorithm needs multiple iterations to evolve current solutions to better solutions for the objective function. This paper explores the possibility of integrating artificial neural network with genetic algorithm to reduce the computation time and improve the performance of this meta-heuristic. In the proposed technique, artificial neural network performs as a brain for the search engine that tries to learn the importance and relations between input and output variables and guide the genetic algorithm for the better solutions. This is implemented by modifying two best solutions of each population using the information stored in artificial neural network. These new solutions are replaced with two worst solutions and get chance to participate in crossover process. The proposed technique is tested using common benchmark problems and the results showed that this hybrid algorithm can find the better solutions in shorter time compared to the traditional genetic algorithm.