Weighted Similarity Classifier Using Differential Evolution and Genetic Algorithm in Weight Optimization
Pasi Luukka, and Jouni Sampo
Laboratory of Applied Mathematics, Lappeenranta University of Technology, P.O. Box 20, FIN-53851 Lappeenranta, Finland
We have compared the differential evolution and genetic algorithms in a study of weight optimization for different similarity measures in a task of classification. In a study of high dimensional data weighting similarity measures become of great importance and efforts to study suitable optimizers is needed. In this article we have studied proper weighting of similarity measures in the classification of high dimensional and large scale data. We will show that in most cases the differential evolution algorithm should be used in finding the weights instead of the genetic algorithm.