Super RaSE: Super Random Subspace Ensemble Classification

We propose a new ensemble classification algorithm, named super random subspace (Super RaSE), to tackle the sparse problem. The proposed algorithm is motivated by (RaSE). RaSE method was shown be flexible framework that can coupled with any existing base classification. However, success of largely depends on proper choice classifier, which unfortunately unknown us. In this work, we show Super avoids need choose classifier randomly sampling collection classifiers together subspace. As result, more and robust than RaSE. addition vanilla RaSE, also develop iterative adaptively changes distribution as well distribution. its version perform competitively for wide range simulated data sets two real examples. are implemented in R package RaSEn.