Learn Before
Activity (Process)

Constructing a Validation-Error Learning Curve from Training-Set Subsets

Choose several increasing training-set sizes and train a separate copy of the same learning algorithm on a subset of each size. Evaluate every trained model on the same development set, then plot the number of training examples on the x-axis and development-set error on the y-axis. For example, from a pool of 1,000 labeled examples, models could be trained on subsets of 100, 200, 300, and progressively larger sizes through 1,000 examples.

0

1

Updated 2026-08-30

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