semi-supervised

Definition

Learning from a small labelled set plus a large pool of unlabelled examples, typically via pseudo-labelling or consistency regularization that extracts signal from the unlabelled part. It targets domains where annotation is the expensive bottleneck. It sits between supervised learning, where every example carries a label, and self-supervised learning, which uses none and invents its own objective.

Explainers using this term

• Graph Neural Networks from Scratch — Learning from ConnectionsSemi-Supervised Classification with Graph Convolutional Networks