Exploring Conflicting Objectives with MADNS: Multiple Assessment Directed Novelty Search

Davy Smith, Laurissa Tokarchuk and Geraint Wiggins

Abstract

Novelty search is an evolutionary approach which promotes phenotypic diversity in a population. Novelty search has been successfully applied to a wide range of domains and a number of variants have been proposed. Here we introduce Multiple Assessment Directed Novelty Search (MADNS), which exploits the notion that a diverse population optimised through phenotypic novelty may contain solutions to multiple conflicting objectives. We show that by utilising the MADNS algorithm, an evolutionary trajectory may be simultaneously directed towards conflicting objectives. We conclude that, through applying MADNS and MC-MADNS, a divergent evolutionary trajectory may be directed to provide simultaneous solutions to multiple conflicting problems in domains with large potential for exploration.

Open Full Paper