Beginner · Data

Validation Data

Visual diagram · (in preparation) · Math · (in preparation) · Worked example · 3 difficulty levels.

TL;DR. A held-out subset used to tune hyperparameters and detect overfitting during training.

Technical Definition

A held-out subset used to tune hyperparameters and detect overfitting during training.

How it works

After each epoch, the model is evaluated on the validation set. Performance there guides choices like early stopping, learning rate, and architecture. The validation set must never be used to update model weights, only to make decisions about them.

Related Concepts

• Cross-Validation — A technique that evaluates model performance by training and testing on different subsets of the data in rotation.
• Training Data — The portion of a dataset used to fit a model's parameters.
• Test Data — A separate, untouched dataset used only at the end to estimate real-world performance.