Generalization Error in Machine Learning (2026 Guide)
Updated on February 01, 2026 6 minutes read
Frequently Asked Questions
What is generalization error in machine learning?
Generalization error is how much a model’s performance drops when you move from training data to new, unseen data. It’s often discussed as test error or the generalization gap.
How do I know if my model is overfitting or underfitting?
Compare training and validation results. Underfitting usually shows high error on both; overfitting often shows low training error but much higher validation/test error.
Does the bias–variance decomposition always apply?
The clean noise + bias² + variance decomposition is exact for squared-error regression under standard assumptions. For classification, the intuition still helps, but the math differs.
What’s a quick way to reduce variance without changing the dataset?
Try regularization (like L2/weight decay), early stopping, or a simpler model. Cross-validation can also help you pick settings that generalize better.