Learn Before
Concept icon
Concept

Using a Smaller Model to Control Variance

Reducing the number of layers or units can make a model less sensitive to the training set, but it also lowers capacity and can increase bias. When extra computation is acceptable, regularization is usually the better first choice because it often improves classification results without forcing the model to be smaller. A compact model is most useful when faster training or lower inference cost matters.

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