Orthogonal Evolution of Teams: A Class of Algorithms for Evolving Teams with Inversely Correlated Errors

Abstract

Several general evolutionary approaches have provenquite successful at evolving teams (or ensembles)consisting of cooperating team members. However, inthis paper we demonstrate that the existing approacheshave subtle, but significant, weaknesses. We thenpresent a novel class of evolutionary algorithms(orthogonal evolution of teams (OET)) for evolvingteams that overcomes these weaknesses. Specifically itis shown that a typical algorithm from the OET class ofalgorithms successfully generates team members thathave fitnesses comparable to those evolvedindependently and that have inversely correlatederrors, which maximises the teams' overall performance.Finally it is shown that the OET approach performssignificantly better than the standard evolutionaryapproaches.