Overfitting vs Underfitting: How to Tell Which One Your Model Has
Underfitting and overfitting, diagnosed from training and validation error, with a polynomial example, a learning-curve check and the fix for each.
Explainers for people learning machine learning. Each one answers a single question, links to the reports, learning paths and playground tools that go deeper, and cites the primary papers behind its claims.
Underfitting and overfitting, diagnosed from training and validation error, with a polynomial example, a learning-curve check and the fix for each.
How a network's loss, its gradients and the learning rate fit together, with a three-step worked example you can check by hand.
Each explainer is written for this site and is not peer-reviewed research. The claims link to primary sources, and the reports they point to list their own sources.