arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2605.08280cs.LGcs.AI
收藏

超越虚假的权衡:适应性EWC用于隐蔽且可泛化的T2I后门

Beyond the False Trade-off: Adaptive EWC for Stealthy and Generalizable T2I Backdoors

Monash University(墨尔本大学)

Lu Bowen, Xinyu Tang, Yin Yin Low, Shu-Min Leong

AI总结:

本文提出Cosine-Aware Adaptive EWC,通过动态调整EWC正则化提升T2I后门攻击的ASR与模型保真度平衡,增强对域外数据集的鲁棒性。

摘要:

在保持模型保真度方面,现有方法如学习不遗忘(LwF)依赖输出基于的知识蒸馏,其正则化效果有限。我们引入弹性权重巩固(EWC)作为参数基于的替代方法,以保持后门学习中的保真度。尽管在理论上更强,我们发现标准静态EWC使用固定正则化权重lambda和均方效用损失会引入攻击成功率(ASR)和保真度之间的虚假权衡,尤其在弱触发器上表现不佳。为此,我们提出Cosine-Aware Adaptive EWC,通过使用基于余弦的语义效用和自适应调度动态调整EWC正则化。此方法将EWC从固定惩罚转变为上下文敏感的约束,从而在保持高ASR的同时维持模型保真度。实验表明,与现有基线相比,该方法在ASR-保真度平衡和增强对域外数据集的鲁棒性方面表现更优。

英文摘要:

Preserving model fidelity is essential for stealthy text-to-image (T2I) backdoor attacks. Existing methods such as Learning without Forgetting (LwF) rely on output-based distillation, which provides limited regularization. We introduce Elastic Weight Consolidation (EWC) as a parameter-based alternative for preserving fidelity in backdoor learning. While stronger in principle, we show that standard static EWC with a fixed regularization weight lambda and mean-squared utility loss creates an artificial trade-off between attack success rate (ASR) and fidelity, particularly degrading performance on weak triggers. To address this, we propose Cosine-Aware Adaptive EWC, which dynamically adjusts EWC regularization using a cosine-based semantic utility and adaptive scheduling. This approach transforms EWC from a fixed penalty into a context-sensitive constraint, maintaining high ASR while preserving model fidelity. Experiments demonstrate improved ASR-fidelity balance and enhanced robustness on out-of-domain (OOD) datasets compared to existing baselines.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑