An Efficient PSO-GA Based Back Propagation Learning-MLP (PSO-GA-BP-MLP) for Classification

In last few decades, Evolutionary computation and Swarm intelligence are two hot favorites for almost all types of researchers. Moreover, many contributions have been made in two directions: Genetic Algorithm (GA) and Particle Swarm optimization (PSO). But, some limitations in both the algorithms (complicated operator like crossover and mutation in GA and early convergence in PSO), are the major restricted boundaries for solving complex problems. In this paper, a hybridization of Particle swarm optimization and Genetic algorithm has been proposed with the back propagation learning based Multilayer perceptron neural network. The effectiveness of the proposed algorithm is shown through a no. of simulation steps with the help of the benchmark datasets considered from UCI machine learning repository. The performance of the algorithm is compared with other standard algorithms to show the steadiness and efficiency as well as statically significant.