Supervised learning
Machine learning from labeled input–output pairs: the model learns to map examples to known targets before deployment.
Supervised learning trains a model on labeled pairs: inputs with known correct outputs (classes, scores, or text completions). The optimizer minimizes a loss between predictions and those labels.
It is a core strand of machine learning, distinct from unlabeled clustering and from reinforcement learning driven by rewards. Large-scale pretraining and fine-tuning of LLMs often use supervised objectives on fixed pairs; RLHF adds preference-based stages afterward. Training in product teams may mix supervised, fine-tuning, and RL phases before inference.