Using Pre-Training Can Improve Model Robustness and Uncertainty

Dan Hendrycks, Kimin Lee, Mantas Mazeika
Proceedings of the 36th International Conference on Machine Learning, PMLR 97:2712-2721, 2019.

Abstract

He et al. (2018) have called into question the utility of pre-training by showing that training from scratch can often yield similar performance to pre-training. We show that although pre-training may not improve performance on traditional classification metrics, it improves model robustness and uncertainty estimates. Through extensive experiments on label corruption, class imbalance, adversarial examples, out-of-distribution detection, and confidence calibration, we demonstrate large gains from pre-training and complementary effects with task-specific methods. We show approximately a 10% absolute improvement over the previous state-of-the-art in adversarial robustness. In some cases, using pre-training without task-specific methods also surpasses the state-of-the-art, highlighting the need for pre-training when evaluating future methods on robustness and uncertainty tasks.

Cite this Paper


BibTeX
@InProceedings{pmlr-v97-hendrycks19a, title = {Using Pre-Training Can Improve Model Robustness and Uncertainty}, author = {Hendrycks, Dan and Lee, Kimin and Mazeika, Mantas}, booktitle = {Proceedings of the 36th International Conference on Machine Learning}, pages = {2712--2721}, year = {2019}, editor = {Kamalika Chaudhuri and Ruslan Salakhutdinov}, volume = {97}, series = {Proceedings of Machine Learning Research}, month = {09--15 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v97/hendrycks19a/hendrycks19a.pdf}, url = { http://proceedings.mlr.press/v97/hendrycks19a.html }, abstract = {He et al. (2018) have called into question the utility of pre-training by showing that training from scratch can often yield similar performance to pre-training. We show that although pre-training may not improve performance on traditional classification metrics, it improves model robustness and uncertainty estimates. Through extensive experiments on label corruption, class imbalance, adversarial examples, out-of-distribution detection, and confidence calibration, we demonstrate large gains from pre-training and complementary effects with task-specific methods. We show approximately a 10% absolute improvement over the previous state-of-the-art in adversarial robustness. In some cases, using pre-training without task-specific methods also surpasses the state-of-the-art, highlighting the need for pre-training when evaluating future methods on robustness and uncertainty tasks.} }
Endnote
%0 Conference Paper %T Using Pre-Training Can Improve Model Robustness and Uncertainty %A Dan Hendrycks %A Kimin Lee %A Mantas Mazeika %B Proceedings of the 36th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2019 %E Kamalika Chaudhuri %E Ruslan Salakhutdinov %F pmlr-v97-hendrycks19a %I PMLR %P 2712--2721 %U http://proceedings.mlr.press/v97/hendrycks19a.html %V 97 %X He et al. (2018) have called into question the utility of pre-training by showing that training from scratch can often yield similar performance to pre-training. We show that although pre-training may not improve performance on traditional classification metrics, it improves model robustness and uncertainty estimates. Through extensive experiments on label corruption, class imbalance, adversarial examples, out-of-distribution detection, and confidence calibration, we demonstrate large gains from pre-training and complementary effects with task-specific methods. We show approximately a 10% absolute improvement over the previous state-of-the-art in adversarial robustness. In some cases, using pre-training without task-specific methods also surpasses the state-of-the-art, highlighting the need for pre-training when evaluating future methods on robustness and uncertainty tasks.
APA
Hendrycks, D., Lee, K. & Mazeika, M.. (2019). Using Pre-Training Can Improve Model Robustness and Uncertainty. Proceedings of the 36th International Conference on Machine Learning, in Proceedings of Machine Learning Research 97:2712-2721 Available from http://proceedings.mlr.press/v97/hendrycks19a.html .

Related Material