Abstract
In this paper, we present BENCH (Biclusteringdriven ENsemble of Classi?ers), an algorithm to construct an ensemble of classi?ers through concurrent feature and data point selection guided by unsupervised knowledge obtained from biclustering. BENCH is designed for underdetermined problems. In our experiments, we use Bayesian Belief Network (BBN) classi?ers as base classi?ers in the ensemble; however, BENCH can be applied to other classi?cation models as well. We show that BENCH is able to increase prediction accuracy of a single classi?er and traditional ensemble of classi?ers by up to 15% on three microarray datasets using various weighting schemes for combining individual predictions in the ensemble.