135
click to vote
KDD
2002
ACM
16 years 3 months ago
2002
ACM
In this paper, we present a new co-training strategy that makes use of unlabelled data. It trains two predictors in parallel, with each predictor labelling the unlabelled data for...