Machine Learning Research
Less Labels, More Learning: Improved small data performance with combined techniques
In small data settings where labels are scarce, semi-supervised learning can train models by using a small number of labeled examples and a larger set of unlabeled examples. A new method outperforms earlier techniques.