Outlining a Practitioner’s Guide to Fitness Function Design

Fitness function design is often both a design and performance bottleneck for evolutionary algorithms. The fitness function for a given problem is directly related to the specifications for that problem. In this paper we outline a guide for transforming problem specifications into a fitness function. The target audience for this guide are non-expert practitioners. The goal of the research presented in this paper is to investigate and formalize the fitness function generation process that expert developers go through and in doing so make fitness function design less of a bottleneck. Solution requirements in the problem specifications are identified and classified; then an appropriate fitness function component is generated based on its classifications. The fitness function components are then combined to yield a fitness function for the problem in question. Illustrative examples utilizing the guide are presented. Also, the performance of a guide generated fitness function is compared to that of an expert designed fitness function demonstrating the competitiveness of the guide generated fitness function.