Generalization error bounds in semi-supervised classification under the cluster assumption

Philippe Rigollet

We consider semi-supervised classification when part of the available data is unlabeled. These unlabeled data can be useful for the classification problem when we make an assumptio…