Density Uncertainty Layers for Reliable Uncertainty Estimation

arXiv:2306.12497v2 Announce Type: replace
Abstract: Assessing the predictive uncertainty of deep neural networks is crucial for safety-related applications of deep learning. Although Bayesian deep learning offers a principled framework for estimating model uncertainty, the common approaches that approximate the parameter posterior often fail to deliver reliable estimates of predictive uncertainty. In this paper, we propose a novel criterion for reliable predictive uncertainty: a model's predictive variance should be grounded in the empirical density of the input. That is, the model …

abstract applications arxiv bayesian bayesian deep learning cs.lg deep learning framework networks neural networks paper posterior predictive safety stat.ml type uncertainty