When: Wednesday, September 20, 2017 – 11:00am – 12:00pm
Brock University Computer Science Department Seminar Series
Prof. Andries Engelbrecht
University of Pretoria
Pretoria, South Africa
Heterogeneous Particle Swarm Optimizers: Scalability,
Self-Adaptive Behaviors, and Dynamic Environments
Abstract: Particles in the standard particle swarm optimization (PSO) algorithm, and most of its modifications, follow the same behaviours. That is, particles implement the same velocity and position update rules. This means that particles exhibit the same search characteristics. Heterogeneous PSO (HPSO), on the other hand, allow particles to follow different search behaviours selected from a behaviour pool, thereby efficiently addressing the exploration–exploitation trade-off problem.
This talk will provide an overview of very simple HPSO algorithms, and will provide empirical results to show that they are more efficient than the homogeneous PSO algorithms making up the behaviour pool of the HPSO algorithm. The talk will also discuss the HPSO scalability properties, and will present approaches to self-adaptive behaviour selection. The talk will end with an illustration of the performance of simple HPSO algorithms in dynamic environments.
Professor Andries Engelbrecht is the South African Research Chair of Artificial Intelligence in the Computer Science Department, University of Pretoria
Location of your event: H313