Concept icon
Concept

Testing a Data-Mismatch Hypothesis on a Shared Subset

If both your training set and your training-dev set contain the same special subset of examples, evaluate that subset separately. For example, if both splits include calls recorded in a noisy subway station, compare performance on those clips in each split. If the model performs well on that subset in training but much worse on the same subset in training-dev, that supports the idea that getting more data from that source could improve the system.

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

Learn After