Log-normal Mutations and their Use in Detecting Surreptitious Fake Images
This work addresses the challenge of detecting surreptitious fake images for security and verification applications, presenting an incremental improvement by adapting existing optimization methods.
The paper tackled the problem of adversarial attacks on fake image detectors by using log-normal mutations from black-box optimization, achieving successful attacks that evade detection by specialized detectors. It then combined these attacks with deep detection to create improved fake detectors.
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.