I was reading an article about metric learning lately. http://arxiv.org/pdf/1306.6709.pdf. In the paper, the author indicates that there are three types of metric learning paradigms, i.e., the full supervised learning, the weak supervised learning and the semi-supervised learning.
What I was confused a bit is that, the author mentioned that for the semi-supervised learning, "besides the full or weak supervision, the algorithm has access to a typically large sample of unlabeled instances for which no side information is available. This is typically useful to avoid overfitting when the labeled data or side information is scarce". So, can anyone explain to me why the unlabeled instances would help to avoid overfitting?