Bias-Variance Tradeoff
noun
say it: “BY-us VAIR-ee-unss TRAYD-awf” /ˈbaɪəs ˈvɛriəns ˈtreɪdˌɔf/
The balance in machine learning between a model too simple to capture a pattern and one so complex that it also captures random noise.
What is the bias-variance tradeoff?
The bias-variance tradeoff is the balance between a model too simple to catch the pattern (high bias: underfitting) and one so flexible it learns each example's noise (high variance: overfitting). Making a model more flexible lowers one and raises the other, so the best setting sits in between.
In a sentence
- “The validation curve made the bias-variance tradeoff easy to see.”
- “Picking the tree's depth is the bias-variance tradeoff in action: too shallow misses the pattern, too deep memorizes noise.”
Level: Freelancer · More AI Basics terms at this level
Updated