AI & MLRecommendedAIML-DATA-013
Semi-Supervised Learning and Pseudo-Labeling
Learning from a few labels plus many unlabeled examples: pseudo-labels, consistency regularisation, and the cluster assumption.
- Difficulty
- 4 / 5
- Importance
- 4 / 5
- Study time
- 1–2 h
- Kind of work
- Mixed
Why it matters
The learner can run self-training with a confidence threshold and predict when pseudo-labels help and when they reinforce early mistakes.