Paper abstract

Parameter Control Methods for Selection Operators in Genetic Algorithms

Peter Vajda, Agoston E. Eiben, and Wim Hordijk

In G. Rudolph et al. (eds.), Parallel Problem Solving from Nature -- PPSN X, Springer, 620-630, 2008.

Parameter control is still one of the main challenges in evolutionary computation. This paper is concerned with controlling selection operators on-the-fly. We perform an experimental comparison of such methods on three groups of test functions and conclude that varying selection pressure during a GA run often yields performance benefits, and therefore is a recommended option for designers and users of evolutionary algorithms.