Uncertainty Estimation by Density Aware Evidential Deep Learning

This repository contains the official PyTorch implementation of the paper Uncertainty Estimation by Density Aware Evidential Deep Learning, accepted for publication at ICML 2024.

Abstract

Evidential deep learning (EDL) has shown remarkable success in uncertainty estimation. However, there is still room for improvement, particularly in out-of-distribution (OOD) detection and classification tasks. The limited OOD detection performance of EDL arises from its inability to reflect the distance between a test example and the training data when quantifying uncertainty, while its limited classification performance stems from its parameterization of the concentration parameters. To address these limitations, we propose a novel method called Density Aware Evidential Deep Learning (DAEDL). DAEDL integrates the feature-space density of a test example with the output of EDL during the prediction stage, while using a novel parameterization that resolves the issues in the conventional parameterization. We prove that DAEDL enjoys a number of favorable theoretical properties. DAEDL demonstrates state-of-the-art performance across diverse downstream tasks related to uncertainty estimation and classification.

Method

Citation

If this code or paper has been useful in your research, please consider citing our paper:

@inproceedings{yoon2024uncertainty,
  title={Uncertainty Estimation by Density Aware Evidential Deep Learning},
  author={Yoon, Taeseong and Kim, Heeyoung},
  booktitle={International Conference on Machine Learning},
  pages={57217--57243},
  year={2024},
  organization={PMLR}
}