UCT CS Research Document Archive

Particle Swarm Optimization with Spatially Meaningful Neighbours

Lane, James, Andries Englebrecht and James Gain (2008) Particle Swarm Optimization with Spatially Meaningful Neighbours. In Proceedings IEEE Swarm Intelligence Symposium, Missouri.

Full text available as:
PDF - Requires Adobe Acrobat Reader or other PDF viewer.

Abstract

Neighbourhood topologies in particle swarm optimization (PSO) are typically random in terms of the spatial positions of connected neighbours. This study explores the use of spatially meaningful neighbours for PSO. An approach is designed which uses heuristics to leverage the natural neighbours computed with Delaunay triangulation. The approach is compared to standard PSO sociometries and fitness distance ratio approaches. Although intrinsic properties of Delaunay triangulation limit the practical application of this approach to low dimensions results show that it is a successful particle swarm optimizer.

EPrint Type:Conference Paper
Keywords:Particle Swarm Optimization
Numerical Search
Subjects:G Mathematics of Computing: G.1 NUMERICAL ANALYSIS
ID Code:477
Deposited By:Gain, James
Deposited On:18 September 2008