Comparative study of Genetic Algorithms and resampling methods for ensemble constructing

Diversity and accuracy in the members of the classifier ensemble appear as two of the main issues to take into account for its construction and operation. The resampling method has been the strategy to construct the most used ensembles; however, the subsamples here obtained consider both diversity and high accuracy. In this work two different strategies to construct ensembles with those characteristics are analyzed: resampling methods as bagging and boosting, and an evolutive strategy as genetic algorithms. Using a dynamic weighting scheme, the genetic algorithm strategy demonstrated its effectiveness in searching the best solution to the problem. In addition, we also introduce other modifications in order to reduce the processing time of the genetic algorithm. All of them are studied specifically in the framework of the nearest neighbour classification algorithm.