Uncertainty Estimation by Flexible Evidential Deep Learning

Taeseong Yoon, Heeyoung Kim · 2025

Read Uncertainty Estimation by Flexible Evidential Deep Learning by Taeseong Yoon, Heeyoung Kim (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2510-18322.txt.

Description

Taeseong Yoon and Heeyoung Kim improve uncertainty estimation in a 2025 machine learning paper. Uncertainty quantification means stating how unsure a model is about each prediction. Overconfident errors can cause real harm in high-stakes uses like medicine or finance.

Evidential deep learning estimates uncertainty by predicting a Dirichlet distribution, a standard model for spreads over class probabilities. The authors generalize that distribution into a more flexible form, giving a richer picture of doubt. Theory supports several advantages of the flexible version. Tests across standard, long-tailed, and noisy datasets show top uncertainty performance, including cases where rare classes dominate the difficulty.

Catalog record

CREATORTaeseong Yoon, Heeyoung Kim
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE139 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2510-18322

Browse the full library · About the Tower of Babel