Generalization
A model's ability to perform well on new, unseen data that wasn't part of its training set -- the ultimate goal of machine learning.
Why It Matters
A model that only performs well on training data (overfitting) is useless in production. Generalization means the model has learned underlying patterns rather than memorizing specific examples.
Improving Generalization
More diverse training data, regularization (dropout, weight decay), data augmentation, cross-validation, early stopping, and proper train/validation/test splits. The bias-variance tradeoff frames the fundamental challenge.