Fixed, random or fuzzy adaptive parameters in an evolutionary algorithm

Louis Gacôgne (1)

1 MALIRE - Machine Learning and Information Retrieval

Particular steady-state strategies of evolution with small sized population are studied in this paper. We specially focus our attention on the ways to reach a compromise with the best parameters, we observe improvements in order to optimize classical problems when using small populations and some clearing to make exploration directly connected to the homogeneity of the population We use for this fuzzy rules or random way to decide and move parameters.

Domaines: Informatique [cs]

Louis Gacôgne. Fixed, random or fuzzy adaptive parameters in an evolutionary algorithm. IEEE Workshop On Hybrid Intelligent Models And Applications (HIMA), SSCI, Apr 2011, Paris, France. pp.33-38, ⟨10.1109/HIMA.2011.5953968⟩. ⟨hal-01286229⟩