Niching in evolution strategies

Evolutionary Algorithms (EAs), popular search methods for solving search and optimization problems, are known for successful and fast location of single optimal solutions. However, many complex search problems require the location and maintenance of multiple solutions. Niching methods, the extension of EAs to address this problem, have been investigated up to date mainly within the field of Genetic Algorithms (GAs). In our study we investigate the basis for niching methods within Evolution Strategies (ES), and propose as far as we know the first ES niching method. We analyze our algorithm and present an experimental setup for testing it in optimization tasks, with emphasis on highdimensional functions. Our algorithm is shown to be robust and to fulfill its goals in finding multiple solutions in optimization problems in one run.

• Type: Conference
• Year: 2005
• Where: GECCO
• Authors: Ofer M. Shir, Thomas Bäck
• Keywords: Algorithm, GECCO 2005, Multiple Solutions, Optimization Problems