Reliable Uncertainty Estimation via Discriminative Feature Learning for Evidential Deep Classification
IEEE Transactions on Pattern Analysis and Machine Intelligence · 2026-10 · 48 (10) · 13477-13495
DOI: 10.1109/tpami.2026.3712710 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Pattern Analysis and Machine Intelligence, 2026-10. Verfasst von Linye Li und 5 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 84 Literaturangaben. Gefördert von: National Natural Science Foundation of China.
Abstract (Originalsprache)
Evidential Deep Learning (EDL) has emerged as a widely adopted framework for quantifying uncertainty in classification tasks by modeling a higher-order prior on classification probabilities using a Dirichlet distribution. Recent advances in EDL have notably progressed, primarily by calibrating the predicted Dirichlet distribution to enhance uncertainty estimation. However, these methods often overlook the role of feature representations, the fundamental components of deep learning models that directly impact both classification performance and uncertainty estimation. This paper adopts a feature-centric perspective to explore the relationship between predicted evidence and feature representations in EDL. We demonstrate that existing EDL approaches struggle to learn sufficiently discriminative features as the number of classes increases, which significantly undermines their capacity to produce reliable uncertainty estimates. To tackle this issue, we propose Tangential Gradient Descent (TGD), a novel feature optimization method that explicitly decomposes the loss gradient with respect to feature representations into tangential and radial components. By accelerating updates along the tangential component, TGD promotes feature alignment to a simplex Equiangular Tight Frame (ETF), thereby maximizing inter-class angular separability for enhanced classification. At the same time, updates along the radial component foster discriminative feature projections that effectively differentiate between certain and uncertain samples, leading to improved uncertainty estimation. Extensive experiments on benchmarks such as CIFAR-10, CIFAR-100, and ImageNet-200, covering both OOD detection and generalization across various class scales, reveal that our approach consistently surpasses conventional EDL methods in both uncertainty estimation and scalability.
Förderung
- National Natural Science Foundation of China
Zitieren
APA
Li, L., Chen, Y., Yue, X., Chen, Q., He, L., & Tao Shen, H. (2026). Reliable Uncertainty Estimation via Discriminative Feature Learning for Evidential Deep Classification. IEEE Transactions on Pattern Analysis and Machine Intelligence, 48(10), 13477-13495. https://doi.org/10.1109/tpami.2026.3712710
Vancouver
Li L, Chen Y, Yue X, Chen Q, He L, Tao Shen H. Reliable Uncertainty Estimation via Discriminative Feature Learning for Evidential Deep Classification. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2026;48(10):13477-13495. doi:10.1109/tpami.2026.3712710
BibTeX
@article{Li2026,
author = {Li, Linye and Chen, Yufei and Yue, Xiaodong and Chen, Qunjie and He, Lianghua and Tao Shen, Heng},
title = {Reliable Uncertainty Estimation via Discriminative Feature Learning for Evidential Deep Classification},
journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
year = {2026},
volume = {48},
number = {10},
pages = {13477--13495},
doi = {10.1109/tpami.2026.3712710}
}
EUVIMED