What is overfitting, and how can you recognize it?

Overfitting occurs when a model captures training-specific noise or patterns that do not generalize. A common sign is strong training performance with substantially weaker validation performance on representative, independent data. Check learning curves, leakage, sample size, and whether the validation split resembles deployment. Possible remedies include a simpler model, regularization, early stopping, and more representative data. Evaluate changes on validation data and reserve a final test set. A small training-validation gap alone does not prove success because both scores may be poor.