All terms

Supervised Learning

Training a model on labeled examples where each input is paired with the correct output.

Machine Learning1 min read

Definition

Supervised learning uses labeled datasets: emails marked spam/not spam, images tagged cat/dog, Somali sentences with English translations. The model learns to map inputs to those labels.

It contrasts with unsupervised learning (no labels) and reinforcement learning (reward signals).

In simple terms

Supervised learning is flashcards with answers on the back — the student learns by checking predictions against the known correct side.

Where you see it

  • Fraud detection with labeled transactions.
  • Speech recognition trained on transcribed audio.
  • Fine-tuning LLMs on instruction–response pairs.

How it works

  1. 1.Label data

    Human annotators or heuristics assign targets.

  2. 2.Train

    Minimize loss between predictions and labels.

  3. 3.Evaluate

    Test on held-out labeled data the model never saw.

Why it matters

  • Most production ML classifiers and rankers are supervised — labels are the fuel.

Often confused

  • You always need millions of labels.

    Transfer learning and few-shot prompting reduce label needs; quality beats quantity for many tasks.