Learn Before
Concept icon
Concept

High Training Error Calls for Better Fit, Not More Data

If a model already has high avoidable bias, adding more training examples usually will not solve the problem. Extra data mainly helps reduce variance, while bias changes little. The first priority is to improve performance on the training set; only after that is the model likely to show meaningful gains on dev and test sets.

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