Supervised Learning
A machine learning approach where the model learns from labeled examples — input-output pairs where the correct answer is provided during training.
Why it matters
Most practical ML applications use supervised learning. Understanding it explains why AI systems need curated training data and why data quality matters more than data quantity.
How it works
You give the model thousands of examples where each input is paired with the correct output. A spam classifier sees emails labeled "spam" or "not spam." A translation model sees English sentences paired with their French equivalents. The model adjusts its internal parameters until it can reliably predict the correct output for new, unseen inputs.
Common tasks
- Classification — assign inputs to categories. Spam detection, image recognition, sentiment analysis.
- Regression — predict a continuous value. Price forecasting, risk scoring, demand estimation.
- Sequence-to-sequence — transform one sequence into another. Translation, summarization, code generation.
The labeling bottleneck
Supervised learning needs labeled data, and labeling is expensive. This is why techniques like self-supervised learning (where models create their own labels from raw data) became important — it is how large language models are pre-trained on vast text corpora without anyone manually labeling each example.