AN IMPROVED GREY WOLF OPTIMIZER WITH LVY WALK AS AN OPTIMIZER IN TRAINING MULTI-LAYER PERCEPTRON WITH DECOUPLED NEURAL INTERFACE

Grey Wolf Optimizer (GWO) is inspired by how grey wolves (Canis Lupus) searching its prey. The GWO relatively new swarm-based intelli- gence and the only algorithms that are based on the leadership hierarchy. In GWO, four types of grey wolves such as alpha, beta, delta and omega are employed simulating the leadership hierarchy. Additionally, there are three main steps of hunt- ing, searching for prey, encircling prey and at- tacking prey are implemented. To improve the GWO search ability, this study proposed Lvy - GWO based on Lvy walk. Five well define bench- mark functions were selected in this study. The five benchmark functions were selected based on its features that have many local minima. The results indicate that Lvy -GWO did improve the original GWO based on the error value. Based on Lvy - GWO algorithm. It will be then pro- posed serving as an optimizer in training multi- layer perceptron (MLP) with Decouple Neural In- terface (DNI).