Wasserstein Dropout
Abstract
Despite of its importance for safe machine learning, uncertainty quantification for neural networks is far from being solved. State-of-the-art approaches to estimate neural uncertainties are often hybrid, combining parametric models with explicit or implicit (dropout-based) ensembling. We take another pathway and propose a novel approach to uncertainty quantification for regression tasks, Wasserstein dropout, that is purely non-parametric. Technically, it captures aleatoric uncertainty by means of dropout-based sub-network distributions. This is accomplished by a new objective which minimizes the Wasserstein distance between the label distribution and the model distribution. An extensive empirical analysis shows that Wasserstein dropout outperforms state-of-the-art methods, on vanilla test data as well as under distributional shift, in terms of producing more accurate and stable uncertainty estimates.
Cite
@article{arxiv.2012.12687,
title = {Wasserstein Dropout},
author = {Joachim Sicking and Maram Akila and Maximilian Pintz and Tim Wirtz and Asja Fischer and Stefan Wrobel},
journal= {arXiv preprint arXiv:2012.12687},
year = {2021}
}