Improved Group Search Optimizer based on cooperation among groups for feedforward networks training with Weight Decay
D. N. G. Silva, Luciano D. S. Pacífico, Teresa B. Ludermir · 2011
Training artificial neural networks (ANNs) is a complex task of great importance in problems of supervised learning. Evolutionary algorithms (EAs) are widely used as global searching techniques for optimization in scientific and engineering problems, and these approaches have been introduced to ANNs to perform various tasks, such as connection weight training and architecture design. Recently, a novel optimization algorithm, called Group Search Optimizer (GSO), was introduced, which is inspired by animal searching behaviour an group living theory. In this paper we introduce two hybrid cooperative GSO approaches based on divide-and-conquer paradigm, employing cooperative behaviour among multiple GSO groups to improve the performance of standard GSO. We also applied the Weight Decay (WD) strategy to enhance the generalization power of networks. Experimental results show that our GSO approaches using cooperation are able to achieve better generalization performance than Levenberg-Marquardt (LM) traditional GSO in real benchmark datasets.