Robust Uncertainty Estimation under Distribution Shift via Difference Reconstruction

TL;DR

This paper introduces DRUE, a novel uncertainty estimation method that reconstructs inputs from intermediate layers to better detect distribution shifts, improving reliability in high-stakes applications like medical imaging.

Contribution

The paper proposes Difference Reconstruction Uncertainty Estimation (DRUE), a new approach that reconstructs inputs from intermediate layers to more accurately measure uncertainty under distribution shifts.

Findings

DRUE outperforms existing methods in OOD detection metrics.

DRUE demonstrates robustness across multiple OOD datasets.

The method enhances model reliability in uncertain environments.

Abstract

Estimating uncertainty in deep learning models is critical for reliable decision-making in high-stakes applications such as medical imaging. Prior research has established that the difference between an input sample and its reconstructed version produced by an auxiliary model can serve as a useful proxy for uncertainty. However, directly comparing reconstructions with the original input is degraded by information loss and sensitivity to superficial details, which limits its effectiveness. In this work, we propose Difference Reconstruction Uncertainty Estimation (DRUE), a method that mitigates this limitation by reconstructing inputs from two intermediate layers and measuring the discrepancy between their outputs as the uncertainty score. To evaluate uncertainty estimation in practice, we follow the widely used out-of-distribution (OOD) detection paradigm, where in-distribution (ID) training data are compared against datasets with increasing domain shift. Using glaucoma detection as the ID task, we demonstrate that DRUE consistently achieves superior AUC and AUPR across multiple OOD datasets, highlighting its robustness and reliability under distribution shift. This work provides a principled and effective framework for enhancing model reliability in uncertain environments.