Heuristic search-based stacking of classifiers

Abstract

Currently, the combination of several classifiers is one of the most activefields within inductive learning. Examples of such techniques are boost-ing, bagging and stacking. From these three techniques, stacking isperhaps the least used one. One of the main reasons for this relates to thedifficulty to define and parameterize its components: selecting whichcombination of base classifiers to use, and which classifiers to use as themeta-classifier. The approach we present in this chapter poses thisproblem as an optimization task, and then uses optimization techniquesbased on heuristic search to solve it. In particular, we apply geneticalgorithms to automatically obtain the ideal combination of learningmethods for the stacking system.