Noise pressure: systematic overestimation of population fitness in genetic algorithms with noisy fitness functions

  • Authors:

  • Timothy Meekhof;Terence Soule

  • Affiliations:

  • University of Idaho, Moscow, ID, USA;University of Idaho, Moscow, ID, USA

  • Venue:
  • Proceedings of the 12th annual conference on Genetic and evolutionary computation
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

In applications of genetic algorithms (GA) to real world problems, we often encounter significant amounts of noise in our fitness functions. We show that the interaction of normally-distributed noise and selection pressure inherently cause overestimation of GA population fitness. We call this inherent fitness overestimation noise pressure. Furthermore, we show that oversampling is not a sufficient technique for eliminating noise pressure.