One of the problems surrounding fuzzy knowledge base generation using genetic algorithms is finding an optimal number of fuzzy sets for each premise. A Genetic algorithm developed by the authors for the automatic generation of fuzzy knowledge bases uses a multi-objective method combining error minimization and simplification. This paper proposes solutions based on cluster analysis and validation indices for the numbers of clusters used in predefining the numbers of fuzzy sets. Two different validation indices as well as a combination of one of these with the multi-objective method are compared to the original multi-objective method on both synthetic and experimental data. Results obtained with the proposed techniques showed a considerable improvement over the multi-objective method on both data sets.
|Title of host publication||Annual Meeting of the North American Fuzzy Information Processing Society (IEEE)|
|Publication status||Published - 2006|
|Event||2006 Annual Meeting of the North American Fuzzy Information Processing Society - Montreal, Canada|
Duration: 3 Jun 2006 → 6 Jun 2006
|Conference||2006 Annual Meeting of the North American Fuzzy Information Processing Society|
|Period||03/06/2006 → 06/06/2006|