Using flocks for solving numerical optimization problems

Abstract

The paper deals with an alternative numerical approach to optimum searching, which is similar to the particle swarm optimizer. The proposed approach uses aggregate motion of creatures in the artificial life based on flocks. Creatures have the social tendency to stick together, and to perform lifelike emergent behaviour, which is based on a few simple, local rules. They differ from most other artificial life flocking (Boids-type) implementations by being attracted by creatures that "sit" on the best found result. Their emergent behaviour produces effective search in the space of all possible solutions. Evolution Strategies, which are often used for solving this kind of problems, were used to compare the quality of searches.