Exploring the landscape of spatial robustness
Name
engstrom19a.pdf
Description
Published version
Size
1.83 MB
Format
Adobe PDF
Checksum (MD5)
cc84cbb053d92fc4d03a03d91e740e2c
Author(s)
Engstrom, Logan G.
Tran, Brandon
Tsipras, Dimitris
Schmidt, Ludwig
Madry, Aleksander
Date Issued
June 2019
Journal
Proceedings of the 36th International Conference on Machine Learning
Publisher
MLResearch Press
Citation
Engstrom, Logan et al. "Exploring the landscape of spatial robustness." Proceedings of the 36th International Conference on Machine Learning, June 2019, Long Beach, California, MLResearch Press, June 2019. © 2019 The Authors
Version
Final published version
Abstract
Copyright 2019 by the author(s). The study of adversarial robustness has so far largely focused on perturbations bound in lvnorms. However, state-of-the-art models turn out to be also vulnerable to other, more natural classes of perturbations such as translations and rotations. In this work, we thoroughly investigate the vulnerability of neural network-based classifiers to rotations and translations. While data augmentation offers relatively small robustness, we use ideas from robust optimization and test-time input aggregation to significantly improve robustness. Finally we find that, in contrast to the ip-norm case, first-order methods cannot reliably find worst-case perturbations. This highlights spatial robustness as a fundamentally different setting requiring additional study.
MIT Department
Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science
Terms of Use
Creative Commons Attribution-Noncommercial-Share Alike
Persistent DSpace Link
DOI of Published Version
http://proceedings.mlr.press/v97/engstrom19a.html
