A tunable model for multi-objective, epistatic, rugged, and neutral fitness landscapes

The fitness landscape of a problem is the relation between the solution candidates and their reproduction probability. In order to understand optimization problems, it is essential to also understand the features of fitness landscapes and their interaction. In this paper we introduce a model problem that allows us to investigate many characteristics of fitness landscapes. Specifically noise, affinity for overfitting, neutrality, epistasis, multi-objectivity, and ruggedness can be independently added, removed, and fine-tuned. With this model, we contribute a useful tool for assessing optimization algorithms and parameter settings. Categories and Subject Descriptors F.2.1 [Analysis of Algorithms and Problem Com

• Type: Conference
• Year: 2008
• Where: GECCO
• Authors: Thomas Weise, Stefan Niemczyk, Hendrik Skubch, Roland Reichle, Kurt Geihs
• Keywords: GECCO 2008, Optimization, Reproduction Probability, Solution Candidates, fitness Landscapes