EPIA'03 - 11th Portuguese Conference on Artificial Intelligence

	Session: December 7, 11:0-12:30, Room A
Title:	SAPPO: A Simple, Adaptive, Predator Prey Optimiser
	Arlindo Silva, Ana Neves, Ernesto Costa
Abstract:	The balance of exploration and exploitation in particle swarm optimisation is closely related to the choice of the algorithm’s parameters. Achieving the right balance is essential for the success of a given optimisation task. This choice is a difficult one, since, for different functions being optimised, the ideal parameter sets can also be very distinct. In this paper, we try to deal with this issue by introducing two new mechanisms in the basic particle swarm optimiser. The first mechanism consists in a predator-prey based strategy to help maintain diversity in the swarm, even in later stages of convergence. The second mechanism is a symbiosis inspired adaptive scheme, where a new swarm of particles encoding the algorithm’s parameters is introduced. Each parameter particle is symbiotically linked with a solution particle, thus allowing the co-optimisation of the algorithm’s parameters and the problem’s solutions.

Back to schedule.