Projected Estimators for Robust Semi-supervised Classification

Jesse H. Krijthe, Marco Loog

For semi-supervised techniques to be applied safely in practice we at least want methods to outperform their supervised counterparts. We study this question for classification usin…