# PPD: Permutation Phase Defense Against Adversarial Examples in Deep Learning

## TL;DR

Permutation Phase Defense (PPD) enhances neural network robustness against adversarial attacks by combining image permutation with Fourier phase analysis, achieving state-of-the-art results on MNIST and CIFAR-10.

## Contribution

PPD introduces a novel adversarial defense method using permutation and Fourier phase, inspired by cryptography principles, to improve robustness.

## Key findings

- Achieved state-of-the-art robustness on MNIST and CIFAR-10.
- Effectively resists powerful adversarial attacks.
- Relies on key safekeeping for security, similar to cryptography.

## Abstract

Deep neural networks have demonstrated cutting edge performance on various tasks including classification. However, it is well known that adversarially designed imperceptible perturbation of the input can mislead advanced classifiers. In this paper, Permutation Phase Defense (PPD), is proposed as a novel method to resist adversarial attacks. PPD combines random permutation of the image with phase component of its Fourier transform. The basic idea behind this approach is to turn adversarial defense problems analogously into symmetric cryptography, which relies solely on safekeeping of the keys for security. In PPD, safe keeping of the selected permutation ensures effectiveness against adversarial attacks. Testing PPD on MNIST and CIFAR-10 datasets yielded state-of-the-art robustness against the most powerful adversarial attacks currently available.