20. Self-Supervised Learning

A type of unsupervised learning where the data itself provides the supervision, often by predicting parts of the data from other parts, commonly used in training large language models.