Abstract | ||
---|---|---|
Particle Swarm Optimisation (PSO) implementations are commonly ad hoc creations, despite the high degree of similarity between variants of the algorithm. Although a “canonical” form of the algorithm is generally understood, differences in the execution of the experiment may produce unique results. It is valuable to establish a common understanding of the informational representation and execution of the algorithm, for the purposes of experiment consistency, repeatability and communication. This paper formulates a generalised model for computational expression of the algorithm. An encoding scheme and protocol are presented, which have been derived from a data taxonomy. The model is shown to accommodate a number of disparate variants, representing a range of interests in PSO study. It is demonstrated that, despite conceptual differences, there is much similarity amongst them. This has wide implications regarding the rigour of experimental practice, and validity of variant performance comparison. |
Year | DOI | Venue |
---|---|---|
2012 | 10.1109/CEC.2012.6252967 | IEEE Congress on Evolutionary Computation |
Keywords | Field | DocType |
encoding,particle swarm optimisation,protocols,PSO,ad hoc creations,consolidated model,data taxonomy,encoding scheme,informational representation,particle swarm optimisation variants,protocol | Particle swarm optimization,Data mining,Data modeling,Mathematical optimization,Degree of similarity,Rigour,Computer science,Theoretical computer science,Implementation,Artificial intelligence,Machine learning,Encoding (memory) | Conference |
ISBN | Citations | PageRank |
978-1-4673-1508-1 | 1 | 0.38 |
References | Authors | |
11 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Shannon S. Pace | 1 | 1 | 0.38 |
Andrew Cain | 2 | 7 | 5.79 |
Clinton J. Woodward | 3 | 16 | 4.41 |