Bio-Inspired Adversarial Attack Against Deep Neural Networks

TL;DR

This paper introduces a novel bio-inspired physical adversarial attack on deep neural networks using moving objects, demonstrating how superimposed patterns and motion can deceive DNNs and cause targeted misclassification or detection failure.

Contribution

It presents the first physical attack method involving moving objects, expanding adversarial strategies beyond stationary perturbations and digital inputs.

Findings

Superimposing patterns on moving objects confuses DNNs.

Motion can reduce frame dependency, causing object detection failures.

Physical attacks with moving objects are feasible and effective.

Abstract

The paper develops a new adversarial attack against deep neural networks (DNN), based on applying bio-inspired design to moving physical objects. To the best of our knowledge, this is the first work to introduce physical attacks with a moving object. Instead of following the dominating attack strategy in the existing literature, i.e., to introduce minor perturbations to a digital input or a stationary physical object, we show two new successful attack strategies in this paper. We show by superimposing several patterns onto one physical object, a DNN becomes confused and picks one of the patterns to assign a class label. Our experiment with three flapping wing robots demonstrates the possibility of developing an adversarial camouflage to cause a targeted mistake by DNN. We also show certain motion can reduce the dependency among consecutive frames in a video and make an object detector "blind", i.e., not able to detect an object exists in the video. Hence in a successful physical attack against DNN, targeted motion against the system should also be considered.