Understanding Adversarial Robustness Through Loss Landscape Geometries
Abstract
The pursuit of explaining and improving generalization in deep learning has elicited efforts both in regularization techniques as well as visualization techniques of the loss surface geometry. The latter is related to the intuition prevalent in the community that flatter local optima leads to lower generalization error. In this paper, we harness the state-of-the-art "filter normalization" technique of loss-surface visualization to qualitatively understand the consequences of using adversarial training data augmentation as the explicit regularization technique of choice. Much to our surprise, we discover that this oft deployed adversarial augmentation technique does not actually result in "flatter" loss-landscapes, which requires rethinking adversarial training generalization, and the relationship between generalization and loss landscapes geometries.
Cite
@article{arxiv.1907.09061,
title = {Understanding Adversarial Robustness Through Loss Landscape Geometries},
author = {Vinay Uday Prabhu and Dian Ang Yap and Joyce Xu and John Whaley},
journal= {arXiv preprint arXiv:1907.09061},
year = {2019}
}
Comments
Presented at the ICML 2019 Workshop on Uncertainty and Robustness in Deep Learning, and CVPR 2019 Workshop on The Bright and Dark Sides of Computer Vision: Challenges and Opportunities for Privacy and Security (CV-COPS)