Semi-supervised learning
Learning from a small labeled set plus a large pile of unlabeled data.
Think of it like
Like a teacher grading a few papers in detail, then trusting students to learn the rest from patterns in the unmarked stack.
Example
You label 1,000 medical images and let the model exploit 100,000 unlabeled ones to sharpen its understanding.
How it actually works
Semi-supervised learning bridges supervised and unsupervised: a little labeled data anchors the task while abundant unlabeled data reveals structure that improves generalization. Techniques include pseudo-labeling and consistency training. It shines when labels are scarce or costly but raw data is plentiful — a very common real-world situation.
For product teams
The pragmatic middle path when labeling everything is too expensive but you have plenty of raw data.
For engineers
Combines limited labels with large unlabeled data (pseudo-labels/consistency) to improve over supervised-only.
Related
- Supervised learning — Blends supervised learning.
- Unsupervised learning — Blended with unsupervised learning.
Read anything AI without the jargon
Look up any term in plain English, or save terms as you read with the free Chrome extension.
Open DecoderAdd to Chrome