Concept icon
Concept

How Human Performance Guides Machine Learning Work

A strong human benchmark is useful in several ways. It helps supply labeled examples for tasks people can judge reliably, it supports error analysis because humans can often explain patterns a model misses, and it gives a reference point for estimating the best achievable error. That reference helps a team set a realistic target, and if the model is still far from it, the remaining gap suggests there is substantial room to improve.

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