ADVREPAIR:Provable Repair of Adversarial Attack

arXiv:2404.01642v1 Announce Type: new
Abstract: Deep neural networks (DNNs) are increasingly deployed in safety-critical domains, but their vulnerability to adversarial attacks poses serious safety risks. Existing neuron-level methods using limited data lack efficacy in fixing adversaries due to the inherent complexity of adversarial attack mechanisms, while adversarial training, leveraging a large number of adversarial samples to enhance robustness, lacks provability. In this paper, we propose ADVREPAIR, a novel approach for provable repair of adversarial attacks using limited data. By utilizing …

abstract adversarial adversarial attacks adversarial training arxiv attacks complexity cs.cr cs.lg data domains networks neural networks neuron repair risks safety safety-critical samples training type vulnerability