Overfitting

What Overfitting is

An overfitted model looks excellent on data it has seen and mediocre on anything new. The opposite failure, underfitting, means the model is too simple to capture the real signal.

How it works

It is detected by a widening gap between training and validation performance, and mitigated with more data, regularisation, dropout, early stopping and simpler architectures.

Why it matters

Every practitioner meets it, and the bias-variance trade-off it illustrates guides most model selection decisions.

Common uses

• →Model selection
• →Regularisation tuning
• →Early stopping decisions