Don't fear the unlabelled: safe semi-supervised learning via debiasing.

Hugo Schmutz, Olivier Humbert, Pierre-Alexandre Mattei

Browse the full ICLR paper archive.