Rethinking Aleatoric and Epistemic Uncertainty
Abstract
The ideas of aleatoric and epistemic uncertainty are widely used to reason about the probabilistic predictions of machine-learning models. We identify incoherence in existing discussions of these ideas and suggest this stems from the aleatoric-epistemic view being insufficiently expressive to capture all of the distinct quantities that researchers are interested in. To explain and address this we derive a simple delineation of different model-based uncertainties and the data-generating processes associated with training and evaluation. Using this in place of the aleatoric-epistemic view could produce clearer discourse as the field moves forward.
Cite
Text
Smith et al. "Rethinking Aleatoric and Epistemic Uncertainty." NeurIPS 2024 Workshops: BDU, 2024.Markdown
[Smith et al. "Rethinking Aleatoric and Epistemic Uncertainty." NeurIPS 2024 Workshops: BDU, 2024.](https://mlanthology.org/neuripsw/2024/smith2024neuripsw-rethinking/)BibTeX
@inproceedings{smith2024neuripsw-rethinking,
title = {{Rethinking Aleatoric and Epistemic Uncertainty}},
author = {Smith, Freddie Bickford and Kossen, Jannik and Trollope, Eleanor and van der Wilk, Mark and Foster, Adam and Rainforth, Tom},
booktitle = {NeurIPS 2024 Workshops: BDU},
year = {2024},
url = {https://mlanthology.org/neuripsw/2024/smith2024neuripsw-rethinking/}
}