AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Vicinal Label Supervision for Reliable Aleatoric and Epistemic Uncertainty Estimation

Linye Li, Yufei Chen, Xiaodong Yue

NeurIPS 2025 · Label-efficient and robust learning › Recognition and long-tail · approach: Confidence calibration objectives · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes vicinal label supervision for evidential deep learning to improve joint aleatoric and epistemic uncertainty estimation, replacing level-0 hard labels with vicinity-based soft supervision.

Problem

Uncertainty estimation in safety-critical ML; existing Evidential Deep Learning (EDL) methods rely on level-0 hard labels that supervise the model with full certainty, which is argued to hurt reliable aleatoric and epistemic uncertainty estimation.

Approach

Instead of hard one-hot labels, EDL is supervised with vicinal labels derived from the neighborhood of each training sample, so the Dirichlet distribution reflects both aleatoric and epistemic uncertainty more faithfully. Details beyond the abstract are not available.

Experimental setup

Builds on: Evidential Deep Learning

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy