Research Vocabulary · Level 2 of 5
Out-of-Distribution
Outside the data distribution or conditions assumed by training or evaluation.
The boundary is context-dependent rather than a universal label.
Example
A model trained on daytime photos receives infrared night images.
Listen to the definition and example
Audio transcript
Out-of-Distribution. Outside the data distribution or conditions assumed by training or evaluation. The boundary is context-dependent rather than a universal label. For example: A model trained on daytime photos receives infrared night images.
Explore this concept
Why it matters
This helps you read papers critically and distinguish evidence from claims.
Start with
Related concepts
Quick recall question
Try answering before looking back at the definition.