Concept icon
Concept

When Increasing Capacity Helps One Error and Hurts the Other

In many models, a change that makes the training fit more closely can also make the model less stable on new data. Adding more units, more layers, or extra input features often lowers bias because the model can represent a more complex pattern. The tradeoff is that variance may rise, which means the model can become more sensitive to quirks in the training set. Regularization usually pushes in the opposite direction: it often raises bias a little while reducing variance by limiting overfitting.

0

1

Concept icon
Updated 2026-08-12

Contributors are:

Who are from:

Tags

Machine Learning

Deep Learning

Supervised Learning

Dive into Deep Learning @ D2L

Data Science

Machine Learning Strategy

Machine Learning Yearning @ DeepLearning.AI

Related