WIT Press


Improvement Of Generation Change On SSE Algorithm

Price

Free (open access)

Volume

37

Pages

8

Published

2006

Size

405 kb

Paper DOI

10.2495/DATA060451

Copyright

WIT Press

Author(s)

T. Maruyama & E. Kita

Abstract

The Stochastic Schemata Exploiter (SSE) is one of the evolutionary optimization algorithms for solving the combinatorial optimization problems. For improving the search performance of SSE, we will present the cross generational elitist selection SSE (cSSE) algorithms of which the generation alternation algorithm is improved by using cross generational elitist selection. The SSE and the cSSE are compared with the Minimal Generation Gap (MGG) and the Bayesian Optimization Algorithm (BOA) on deception and knapsack problems in order to discuss their features. 1 Introduction The actual industrial problems are often formulated as the multi-objective optimization problems defined with many design variables. Since it is very time-consuming to find the solutions exactly, we have to obtain the quasi-optimum solutions in an admissible time. For this purpose, the evolutionary computations (ECs) are very useful tool [1–4]. Stochastic Schemata Exploiter (SSE) was presented by Aizawa [5] in 1994. The SSE is one of the EC and effective for the 0/1 combinational optimization problems. In the Genetic Algorithm(GAs), the individuals are generated randomly and the new ones are generated by the genetic operations such as the mutation, the crossover, the selection and so on. In the SSE, the other operations than the mutation are not necessary. Therefore, the total number of the control parameter of the SSE is much smaller than that of original GAs. Sub-populations are generated from the whole population according to the fitness ranking of the individuals. Schemata are extracted from the sub-population. New individuals are generated

Keywords