The Effects of Locality on the Dynamics of Decoder-Based Evolutionary Search

The success of decoder-based evolutionary algorithms (EAs) strongly depends on the achieved locality of operators and decoders. Most approaches to investigate locality properties are static and consider only parts of the complex interactions within an EA, and sometimes, these techniques give misleading results. We suggest an explicit analysis of the dynamic behavior, emphasizing the effects of locality on evolutionary search. The impact of our methodology is twofold since it confirms previous statically obtained results and allows to gain reliable additional insight about the actual dynamics. The approach is successfully applied to four EAs for the multidimensional knapsack problem, and it can easily be adapted to other problems and EAs.

• Type: Conference
• Year: 2000
• Where: GECCO
• Authors: Jens Gottlieb, Günther R. Raidl
• Keywords: Decoder-based Evolutionary Algorithms, GECCO 2000, Locality Properties, Multidimensional Knapsack Problem, Optimization