Adversarial Amendment is the Only Force Capable of Transforming an Enemy into a Friend

TL;DR

This paper proposes adversarial amendment (AdvAmd), a novel method that uses adversarial attacks to enhance neural network accuracy on benign samples, challenging the view of adversarial attacks solely as threats.

Contribution

The paper introduces AdvAmd, a new approach that leverages adversarial attacks to improve neural model accuracy, unlike traditional defense methods.

Findings

AdvAmd heals accuracy degradation on benign samples

AdvAmd boosts accuracy in classification, detection, segmentation

Key components include mediate samples, auxiliary batch norm, and specialized loss

Abstract

Adversarial attack is commonly regarded as a huge threat to neural networks because of misleading behavior. This paper presents an opposite perspective: adversarial attacks can be harnessed to improve neural models if amended correctly. Unlike traditional adversarial defense or adversarial training schemes that aim to improve the adversarial robustness, the proposed adversarial amendment (AdvAmd) method aims to improve the original accuracy level of neural models on benign samples. We thoroughly analyze the distribution mismatch between the benign and adversarial samples. This distribution mismatch and the mutual learning mechanism with the same learning ratio applied in prior art defense strategies is the main cause leading the accuracy degradation for benign samples. The proposed AdvAmd is demonstrated to steadily heal the accuracy degradation and even leads to a certain accuracy boost of common neural models on benign classification, object detection, and segmentation tasks. The efficacy of the AdvAmd is contributed by three key components: mediate samples (to reduce the influence of distribution mismatch with a fine-grained amendment), auxiliary batch norm (to solve the mutual learning mechanism and the smoother judgment surface), and AdvAmd loss (to adjust the learning ratios according to different attack vulnerabilities) through quantitative and ablation experiments.