Local search algorithms for memetic algorithms: understanding behaviors using biological intelligence

Abstract

Memetic Algorithms (MAs) are a class of stochastic global search heuristics in which Evolutionary Algorithms (EAs) - based approaches are combined usually with heuristic local searches. This hybridization is meant to reach solutions that would otherwise be unreachable by evolution or a local method alone. In this work, we propose three Local Search (LS) algorithms for hybridization with an existing Evolutionary Algorithm with Pareto ranking in order to define biological intelligence using the concepts of useful and utility and therefore to zoom on the basin of attraction of promising realistic solutions. Our experimental results with these memetic algorithms in the game of Checkers show how we can learn the organization of behaviors into paths of behaviors of different lengths and frequencies and then reveal the true nature of these behaviors.