# Mitigating large adversarial perturbations on X-MAS (X minus Moving Averaged Samples)

Source: [Mitigating large adversarial perturbations on X-MAS (X minus Moving Averaged Samples)](<https://www.sota2.com/research/paper/68fa66eb1e209131cea132d2>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2025-10-28T13:56:41.501Z

Authors: Woohyung Chun, Sung-Min Hong, Junho Huh, Inyup Kang

Published: 2019-12-19T22:37:12.000Z

## Abstract

We propose the scheme that mitigates the adversarial perturbation $\\epsilon$ on the adversarial example $X\_{adv}$ ($=$ $X$ $\\pm$ $\\epsilon$, $X$ is a benign sample) by subtracting the estimated perturbation $\\hat{\\epsilon}$ from $X$ $+$ $\\epsilon$ and adding $\\hat{\\epsilon}$ to $X$ $-$ $\\epsilon$. The estimated perturbation $\\hat{\\epsilon}$ comes from the difference between $X\_{adv}$ and its moving-averaged outcome $W\_{avg}\*X\_{adv}$ where $W\_{avg}$ is $N \\times N$ moving average kernel that all the coefficients are one. Usually, the adjacent samples of an image are close to each other such that we can let $X$ $\\approx$ $W\_{avg}\*X$ (naming this relation after X-MAS\[X minus Moving Averaged Samples\]). By doing that, we can make the estimated perturbation $\\hat{\\epsilon}$ falls within the range of $\\epsilon$. The scheme is also extended to do the multi-level mitigation by configuring the mitigated adversarial example $X\_{adv}$ $\\pm$ $\\hat{\\epsilon}$ as a new adversarial example to be mitigated. The multi-level mitigation gets $X\_{adv}$ closer to $X$ with a smaller (i.e. mitigated) perturbation than original unmitigated perturbation by setting the moving averaged adversarial sample $W\_{avg} \* X\_{adv}$ (which has the smaller perturbation than $X\_{adv}$ if $X$ $\\approx$ $W\_{avg}\*X$) as the boundary condition that the multi-level mitigation cannot cross over (i.e. decreasing $\\epsilon$ cannot go below and increasing $\\epsilon$ cannot go beyond). With the multi-level mitigation, we can get high prediction accuracies even in the adversarial example having a large perturbation (i.e. $\\epsilon$ $\>$ $16$). The proposed scheme is evaluated with adversarial examples crafted by the FGSM (Fast Gradient Sign Method) based attacks on ResNet-50 trained with ImageNet dataset.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/1912.12170>)

[PDF](<https://arxiv.org/pdf/1912.12170.pdf>)
