AI Dictionary › AI Fundamentals

Early Stopping

Early stopping is a technique that halts a model's training before it reaches the planned maximum number of epochs, when performance on validation data stops improving. It prevents overfitting, the situation where a model keeps improving on training data but its ability to generalize gets worse. It is one of the simplest and most widely used regularization techniques.

Definition

How it works

During training, a metric computed on a validation set separate from the training data is periodically monitored. When this metric stops improving for a certain number of consecutive checks, training is stopped and the checkpoint corresponding to the best-performing point is kept.

Applications

It is common practice when training many types of neural networks, including those used for classification, regression and generation tasks. It is often combined with other regularization techniques such as dropout or weight penalties, to obtain models that generalize better to unseen data.

History & etymology

It is an established regularization technique in machine learning practice, widely adopted for its low implementation cost and its effectiveness in preventing overfitting.

Related terms

More in AI Fundamentals

Put it into practice

From our network

Magellano GPS: Fleet Tracking Made Simple

Real-time GPS tracking, remote engine lock, fuel and CO₂ reporting for your fleet.

Visit magellanogps.com →

From the Agora Intelligence blog

More on agora-intelligence.com →

📱 Download the Android app (beta) iOS coming soon

Say what you mean. Get what you need.

Grace Certified, the AI coach that trains and certifies your prompt engineering, by Agora Intelligence.