Log-normal Mutations and their Use in Detecting Surreptitious Fake Images

TL;DR

This paper explores the use of log-normal mutation algorithms for black-box attacks on fake image detectors, demonstrating their effectiveness and leading to improved detection methods.

Contribution

It introduces the application of log-normal mutation algorithms to generate undetectable fake image attacks and enhances fake detectors by combining these attacks with deep learning.

Findings

Log-normal attacks successfully bypass classical detectors.

Combining attacks with deep detection improves fake image detection.

Log-normal mutation is effective in black-box attack scenarios.

Abstract

In many cases, adversarial attacks are based on specialized algorithms specifically dedicated to attacking automatic image classifiers. These algorithms perform well, thanks to an excellent ad hoc distribution of initial attacks. However, these attacks are easily detected due to their specific initial distribution. We therefore consider other black-box attacks, inspired from generic black-box optimization tools, and in particular the log-normal algorithm. We apply the log-normal method to the attack of fake detectors, and get successful attacks: importantly, these attacks are not detected by detectors specialized on classical adversarial attacks. Then, combining these attacks and deep detection, we create improved fake detectors.