Gray-box Adversarial Training

B. S. Vivek, Konda Reddy Mopuri, R. Venkatesh Babu; Proceedings of the European Conference on Computer Vision (ECCV), 2018, pp. 203-218

Abstract


Adversarial samples are perturbed inputs crafted to mislead the machine learning systems. A training mechanism, called adversarial training, which presents adversarial samples along with clean samples has been introduced to learn robust models. In order to scale adversarial training for large datasets, these perturbations can only be crafted using fast and simple methods (e.g., using gradient ascent). However, it is shown that adversarial training converges to a degenerate minimum, where the model appears to be robust by generating weaker adversaries. As a result, the models are vulnerable to simple black-box attacks. In this paper we, (i) demonstrate the drawbacks of existing evaluation policy, (ii) Introduce novel variants of white-box and black-box attacks, dubbed "gray-box adversarial attacks" based on which we propose novel evaluation method to assess the robustness of the learned models, and (iii) propose a novel variant of adversarial training, named "Graybox Adversarial Training" that uses intermediate versions of the models to seed the adversaries. Experimental evaluation demonstrates that the models trained using our method exhibit better robustness compared to both undefended and adversarially trained models.

Related Material


[pdf]
[bibtex]
@InProceedings{Vivek_2018_ECCV,
author = {Vivek, B. S. and Mopuri, Konda Reddy and Babu, R. Venkatesh},
title = {Gray-box Adversarial Training},
booktitle = {Proceedings of the European Conference on Computer Vision (ECCV)},
month = {September},
year = {2018}
}