Concept icon
Concept

Using Human Labels to Approximate the Best Possible Error

When a task is one that people usually do well, such as identifying common objects in images or writing out short audio recordings, you can sample examples, ask a person to label them, and compare the human labels with the established labels for those examples. The resulting human performance gives a practical estimate of the lowest error rate you should expect on that task.

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