Example

Classifier with 14% Estimated Bias and 16% Estimated Variance

A classifier with estimated bias of 14% retains substantial training error relative to the selected lowest-plausible-error benchmark. Its estimated variance of 16% indicates a further substantial increase in error on held-out data. The classifier therefore shows both poor training fit and poor generalization, so describing it only as underfitting or only as overfitting captures only part of its error pattern.

0

1

Updated 2026-08-30

Contributors are:

Who are from:

Tags

Machine Learning

Deep Learning

Machine Learning Strategy

Supervised Learning

Dive into Deep Learning @ D2L

Data Science

Machine Learning Yearning @ DeepLearning.AI