Variational Randomized Smoothing for Sample-Wise Adversarial Robustness

TL;DR

This paper introduces a variational approach to randomized smoothing that adaptively selects noise levels for each input, improving adversarial robustness and certified guarantees over fixed-noise methods.

Contribution

It proposes a novel variational framework with a per-sample noise level selector, advancing randomized smoothing techniques for better robustness.

Findings

Enhanced empirical robustness against adversarial attacks.

Provides and analyzes certified robustness for the sample-wise smoothing method.

Demonstrates superiority over fixed-noise smoothing approaches.

Abstract

Randomized smoothing is a defensive technique to achieve enhanced robustness against adversarial examples which are small input perturbations that degrade the performance of neural network models. Conventional randomized smoothing adds random noise with a fixed noise level for every input sample to smooth out adversarial perturbations. This paper proposes a new variational framework that uses a per-sample noise level suitable for each input by introducing a noise level selector. Our experimental results demonstrate enhancement of empirical robustness against adversarial attacks. We also provide and analyze the certified robustness for our sample-wise smoothing method.