Towards Robust Memetic Algorithms

Abstract

This chapter reports the results of Multimeme algorithms that employ adaptive helpers. A Multimeme Algorithm resorts to a variety of local search neighborhoods for its local search stage allowing for a more robust global search. Each neighborhood is explored by an adaptive helper that allows no;-improving moves that render the Memetic algorithm even more robust to deceptive local optima. We will report results on the use of a single adaptive helper Memetic algorithm for the Traveling Salesman Problem (TSP) and on adaptive helpers within a Multimeme algorithm for the TSP and Protein Structure Prediction Problem (PSP).