A Taxonomy of Heterogeneity and Dynamics in Particle Swarm Optimisation

Harry Goldingay and Peter R. Lewis
In Lecture Notes in Computer Science 8672 / 2014 "Parallel Problem Solving from Nature - PPSN XIII", pp 171-180. Springer, 2014.

We propose a taxonomy for heterogeneity and dynamics of swarms in PSO, which separates the consideration of homogeneity and heterogeneity from the presence of adaptive and non-adaptive dynamics, both at the particle and swarm level. It thus supports research into the separate and combined contributions of each of these characteristics. An analysis of the literature shows that most recent work has focussed on only parts of the taxonomy. Our results agree with prior work that both heterogeneity and dynamics are useful. However while heterogeneity does typically improve PSO, this is often dominated by the improvement due to dynamics. Adaptive strategies used to generate heterogeneity may end up sacrificing the dynamics which provide the greatest performance increase. We evaluate exemplar strategies for each area of the taxonomy and conclude with recommendations.

@incollection{
year={2014},
isbn={978-3-319-10761-5},
booktitle={Parallel Problem Solving from Nature – PPSN XIII},
volume={8672},
series={Lecture Notes in Computer Science},
editor={Bartz-Beielstein, Thomas and Branke, Jürgen and Filipič, Bogdan and Smith, Jim},
doi={10.1007/978-3-319-10762-2_17},
title={A Taxonomy of Heterogeneity and Dynamics in Particle Swarm Optimisation},
url={http://dx.doi.org/10.1007/978-3-319-10762-2_17},
publisher={Springer International Publishing},
author={Goldingay, Harry and Lewis, Peter R.},
pages={171-180},
language={English}
}