用博弈优化方法高效训练抗多攻击模型，降低计算成本同时提升稳定性。
Stabilizing Multi-Attack Adversarial Training via Bandit Optimization
- 将多攻击对抗训练建模为多臂老虎机问题，每轮只选一个攻击样本
- 在低计算开销下实现比现有方法更强的综合鲁棒性
- 适合需要高效且稳定对抗训练的研究者和工业应用
深度神经网络仍易受多种对抗扰动影响，促使多攻击对抗训练（AT）以提升鲁棒性。然而，现有方法或因每轮计算所有攻击而开销过大，或依赖对抗样本的随机采样，可能引发参数过度漂移。为此，我们提出校准对抗采样（CAS），一种高效稳定的框架，将多攻击对抗训练重新建模为多臂老虎机优化问题。通过每轮动态平衡探索与利用地采样单一攻击，CAS显著降低训练成本，缓解不同攻击间的优化冲突，并有效控制参数漂移。大量实验表明，CAS在极低计算成本下实现了更优的整体鲁棒性，为多攻击场景下的鲁棒泛化提供了可扩展且原理清晰的方法。代码已开源。
原文摘要 · Abstract (English)
Deep Neural Networks (DNNs) remain vulnerable to diverse adversarial perturbations, motivating multi-attack adversarial training (AT) for improved robustness. However, existing methods either incur prohibitive overhead by computing all attacks at each iteration, or rely on stochastic sampling over adversarial examples, which may cause excessive parameter drift. To address these issues, we propose Calibrated Adversarial Sampling (CAS), an efficient and stable framework that reformulates multi-attack AT as a multi-armed bandit optimization problem. By sampling a single attack per iteration that dynamically balances exploration and exploitation, CAS significantly reduces training cost while mitigating optimization conflicts across attacks and controlling excessive parameter drifts. Extensive experiments demonstrate that CAS achieves superior overall robustness at low computational cost, offering a scalable and principled approach to robust generalization against multi-attack settings. Our code is available at https://github.com/1240148048/CAS.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。