December 4, 2025Open Access

Introducing “Inside” out of distribution

Key Points

Machine learning models demonstrate varied performance metrics when faced with inside versus outside out-of-distribution samples.
The F 1 score serves as a critical performance metric, highlighting differences in how models respond to different OOD cases.
Observational analysis across synthetic datasets shows that outside OOD cases generally lead to more significant performance degradation.
These findings underscore the necessity of recognizing both inside and outside OOD profiles for effective counter-OOD strategies.

Abstract

Abstract Detecting and understanding out-of-distribution (OOD) samples is crucial in machine learning (ML) to ensure reliable model performance. Current OOD studies primarily focus on extrapolatory (outside) OOD, neglecting potential cases of interpolatory (inside) OOD. In this study, we introduce a novel perspective on OOD by suggesting that it can be divided into inside and outside cases. We examine the inside–outside OOD profiles of datasets and their impact on ML model performance, using normalized root mean squared error (RMSE) and F 1 score as the performance metrics on synthetically generated datasets with both inside and outside OOD. Our analysis demonstrates that different inside–outside OOD profiles lead to unique effects on ML model performance, with outside OOD generally causing greater performance degradation, on average. These findings highlight the importance of distinguishing between inside and outside OOD for developing effective counter-OOD methods.

Read Full Paperexternally

Mark Helpful

Bookmark

Relay

View Full Paper