Selfish herds optimization algorithm with orthogonal design and information update for training multi-layer perceptron neural network

摘要: Selfish herd optimization algorithm is a novel meta-heuristic algorithm, which simulates the group behavior of herds when attacked by predators in nature. With further research it found that cannot get better global optimal solution solving some problems. In order to improve ability we propose selfish with orthogonal design and information update (OISHO) this paper. Through using method, more competitive candidate can be generated. If than solution, will replace solution. At same time, at end each iteration, population algorithm. The purpose increase diversity population, so expands its search space find solutions. verify effectiveness proposed used train multi-layer perceptron (MLP) neural network. For training network, challenging task present satisfactory effective We chose twenty different datasets from UCI machine learning repository as dataset, experimental results are compared SSA, GG-GSA, GSO, GOA, WOA SOS, respectively. Experimental show has accuracy, convergence speed stability other algorithms for