Self Supervised Learning
In plain terms
Learning from data by generating its own labels.
Going deeper
Self-supervised learning creates training signals from the data itself (e.g., predicting masked words), avoiding manual labeling. It learns from structure within raw data.
Under the hood
Powers modern foundation models; pretext tasks teach representations later fine-tuned for real tasks.
Examples
LLM pretraining, masked language modeling, contrastive learning, embeddings.
Think of it like...
Like learning a language by guessing missing words in sentences you read.