Adapting to the habitat: On the integration of local search into the predator-prey model

Abstract

Traditionally, Predator-Prey Models - although providing a more nature-oriented approach to multi-objective optimization than many other standard Evolutionary Multi-Objective Algorithms - suffer from inherent diversity loss for non-convex problems. Still, the approach to peg single objectives to a predator allows a very simple algorithmic design. The building-block configuration of the predators offers potent means for fine-tuning and tackling multi-objective problems in a problem-specific way. In the work at hand, we propose the integration of local search heuristics into the classic model approach in order to overcome the unsatisfactory behavior for the aforementioned problem class. Our results show that, introducing a gradient-based local search mechanism to the system, deficiencies with respect to diversity loss can be highly ameliorated while keeping the beneficial properties of the Predator-Prey Model. © Springer-Verlag 2009.