Adversarially Robust Learning via Entropic Regularization

TL;DR

This paper introduces ATENT, a novel adversarial training method using entropic regularization to improve robustness of deep neural networks against adversarial attacks, showing competitive results on standard benchmarks.

Contribution

The paper proposes a new loss function with entropic regularization for adversarial training, enhancing robustness by focusing on high-loss regions and neighborhoods in the data space.

Findings

Achieves competitive robust accuracy on MNIST and CIFAR-10.

Introduces a novel loss function with entropic regularization.

Demonstrates improved robustness over existing methods.

Abstract

In this paper we propose a new family of algorithms, ATENT, for training adversarially robust deep neural networks. We formulate a new loss function that is equipped with an additional entropic regularization. Our loss function considers the contribution of adversarial samples that are drawn from a specially designed distribution in the data space that assigns high probability to points with high loss and in the immediate neighborhood of training samples. Our proposed algorithms optimize this loss to seek adversarially robust valleys of the loss landscape. Our approach achieves competitive (or better) performance in terms of robust classification accuracy as compared to several state-of-the-art robust learning approaches on benchmark datasets such as MNIST and CIFAR-10.