Fraunhofer-Gesellschaft

Publica

Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

A Novel Regression Loss for Non-Parametric Uncertainty Optimization

 
: Sicking, Joachim; Akila, Maram; Pintz, Maximilian; Wirtz, Tim; Fischer, Asja; Wrobel, Stefan

:
Volltext urn:nbn:de:0011-n-6347596 (3.6 MByte PDF)
MD5 Fingerprint: 62a070b21ab7988f4d6f6156d4ca8be2
Erstellt am: 11.5.2021


3rd Symposium on Advances in Approximate Bayesian Inference, AABI 2021. Online resource : Virtual Event, January-February, 2021
Online im WWW, 2021
https://openreview.net/group?id=approximateinference.org/AABI/2021/Symposium
27 S.
Symposium on Advances in Approximate Bayesian Inference (AABI) <3, 2021, Online>
Bundesministerium für Bildung und Forschung BMBF (Deutschland)
01S18038B; ML2R
Machine Learning Rhein-Ruhr
Bundesministerium fur Wirtschaft und Energie BMWi (Deutschland)
VDA Leitinitiative autonomes und vernetztes Fahren; 19A19005X
KI Absicherung - Safe AI for Automated Driving
Englisch
Konferenzbeitrag, Elektronische Publikation
Fraunhofer IAIS ()

Abstract
Quantification of uncertainty is one of the most promising approaches to establish safe machine learning. Despite its importance, it is far from being generally solved, especially for neural networks. One of the most commonly used approaches so far is Monte Carlo dropout, which is computationally cheap and easy to apply in practice. However, it can underestimate the uncertainty. We propose a new objective, referred to as second-moment loss (SML), to address this issue. While the full network is encouraged to model the mean, the dropout networks are explicitly used to optimize the model variance. We intensively study the performance of the new objective on various UCI regression datasets. Comparing to the state-of-the-art of deep ensembles, SML leads to comparable prediction accuracies and uncertainty estimates while only requiring a single model. Under distribution shift, we observe moderate improvements. As a side result, we introduce an intuitive Wasserstein distance-based uncertainty measure that is non-saturating and thus allows to resolve quality differences between any two uncertainty estimates.

: http://publica.fraunhofer.de/dokumente/N-634759.html