Multi-K Machine Learning Ensembles

Ensemble machine learning models often surpass single models in classification accuracy at the expense of higher computational requirements during training and execution. In this paper we present a novel ensemble algorithm called Multi-K which uses unsupervised clustering as a form of dataset preprocessing to create component models that lead to effective and efficient ensembles. We also present a modification of Multi-K that we call Multi-KX that incorporates a metalearner to help with ensemble classifications. We compare our algorithms to several existing algorithms in terms of classification accuracy and computational speed.