Evaluating Robustness of Predictive Uncertainty Estimation: Are Dirichlet-based Models Reliable?

Anna-Kathrin Kopetzki, Bertrand Charpentier, Daniel Zügner, Sandhya Giri, Stephan Günnemann
Proceedings of the 38th International Conference on Machine Learning, PMLR 139:5707-5718, 2021.

Abstract

Dirichlet-based uncertainty (DBU) models are a recent and promising class of uncertainty-aware models. DBU models predict the parameters of a Dirichlet distribution to provide fast, high-quality uncertainty estimates alongside with class predictions. In this work, we present the first large-scale, in-depth study of the robustness of DBU models under adversarial attacks. Our results suggest that uncertainty estimates of DBU models are not robust w.r.t. three important tasks: (1) indicating correctly and wrongly classified samples; (2) detecting adversarial examples; and (3) distinguishing between in-distribution (ID) and out-of-distribution (OOD) data. Additionally, we explore the first approaches to make DBU mod- els more robust. While adversarial training has a minor effect, our median smoothing based ap- proach significantly increases robustness of DBU models.

Cite this Paper

BibTeX
@InProceedings{pmlr-v139-kopetzki21a, title = {Evaluating Robustness of Predictive Uncertainty Estimation: Are Dirichlet-based Models Reliable?}, author = {Kopetzki, Anna-Kathrin and Charpentier, Bertrand and Z{\"u}gner, Daniel and Giri, Sandhya and G{\"u}nnemann, Stephan}, booktitle = {Proceedings of the 38th International Conference on Machine Learning}, pages = {5707--5718}, year = {2021}, editor = {Meila, Marina and Zhang, Tong}, volume = {139}, series = {Proceedings of Machine Learning Research}, month = {18--24 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v139/kopetzki21a/kopetzki21a.pdf}, url = {https://proceedings.mlr.press/v139/kopetzki21a.html}, abstract = {Dirichlet-based uncertainty (DBU) models are a recent and promising class of uncertainty-aware models. DBU models predict the parameters of a Dirichlet distribution to provide fast, high-quality uncertainty estimates alongside with class predictions. In this work, we present the first large-scale, in-depth study of the robustness of DBU models under adversarial attacks. Our results suggest that uncertainty estimates of DBU models are not robust w.r.t. three important tasks: (1) indicating correctly and wrongly classified samples; (2) detecting adversarial examples; and (3) distinguishing between in-distribution (ID) and out-of-distribution (OOD) data. Additionally, we explore the first approaches to make DBU mod- els more robust. While adversarial training has a minor effect, our median smoothing based ap- proach significantly increases robustness of DBU models.} }
Endnote
%0 Conference Paper %T Evaluating Robustness of Predictive Uncertainty Estimation: Are Dirichlet-based Models Reliable? %A Anna-Kathrin Kopetzki %A Bertrand Charpentier %A Daniel Zügner %A Sandhya Giri %A Stephan Günnemann %B Proceedings of the 38th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2021 %E Marina Meila %E Tong Zhang %F pmlr-v139-kopetzki21a %I PMLR %P 5707--5718 %U https://proceedings.mlr.press/v139/kopetzki21a.html %V 139 %X Dirichlet-based uncertainty (DBU) models are a recent and promising class of uncertainty-aware models. DBU models predict the parameters of a Dirichlet distribution to provide fast, high-quality uncertainty estimates alongside with class predictions. In this work, we present the first large-scale, in-depth study of the robustness of DBU models under adversarial attacks. Our results suggest that uncertainty estimates of DBU models are not robust w.r.t. three important tasks: (1) indicating correctly and wrongly classified samples; (2) detecting adversarial examples; and (3) distinguishing between in-distribution (ID) and out-of-distribution (OOD) data. Additionally, we explore the first approaches to make DBU mod- els more robust. While adversarial training has a minor effect, our median smoothing based ap- proach significantly increases robustness of DBU models.
APA
Kopetzki, A., Charpentier, B., Zügner, D., Giri, S. & Günnemann, S.. (2021). Evaluating Robustness of Predictive Uncertainty Estimation: Are Dirichlet-based Models Reliable?. Proceedings of the 38th International Conference on Machine Learning, in Proceedings of Machine Learning Research 139:5707-5718 Available from https://proceedings.mlr.press/v139/kopetzki21a.html.

Related Material