Calibrated Adversarial Sampling: Multi-Armed Bandit-Guided Generalization Against Unforeseen Attacks
This addresses safety concerns for AI systems by improving robustness against multiple attack types, though it is incremental as it builds on adversarial training frameworks.
The paper tackles the problem of deep neural networks being vulnerable to unforeseen adversarial attacks by proposing Calibrated Adversarial Sampling (CAS), a fine-tuning method that achieves superior overall robustness while maintaining high clean accuracy on benchmark datasets.
Deep Neural Networks (DNNs) are known to be vulnerable to various adversarial perturbations. To address the safety concerns arising from these vulnerabilities, adversarial training (AT) has emerged as one of the most effective paradigms for enhancing the robustness of DNNs. However, existing AT frameworks primarily focus on a single or a limited set of attack types, leaving DNNs still exposed to attack types that may be encountered in practice but not addressed during training. In this paper, we propose an efficient fine-tuning method called Calibrated Adversarial Sampling (CAS) to address these issues. From the optimization perspective within the multi-armed bandit framework, it dynamically designs rewards and balances exploration and exploitation by considering the dynamic and interdependent characteristics of multiple robustness dimensions. Experiments on benchmark datasets show that CAS achieves superior overall robustness while maintaining high clean accuracy, providing a new paradigm for robust generalization of DNNs.