发表机构
College of Computing and Data Science, Nanyang Technological University（南洋理工大学计算与数据科学学院）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
提出统一参数级框架CSR，在U-Net和Transformer两种架构中通过原生预测空间优化再激活被擦除概念，无需外部数据集，跨架构实验验证其有效性，严格裸体场景下ASR分别达50.47%和40.29%。
AI 中文摘要
概念擦除旨在抑制文本到图像生成模型中的不良知识。然而，现有的鲁棒性评估通常依赖于针对特定模型架构定制的再学习攻击。我们研究了两种截然不同的生成范式中的概念再激活：噪声预测U-Net和流匹配Transformer。我们引入了概念分数再学习（CSR），这是一个统一的参数级框架，通过在各自的原生预测空间内优化每个模型来再激活被擦除的概念。CSR不需要外部目标概念图像数据集，并将相同的概念导向目标应用于基于U-Net的Stable Diffusion和基于Transformer的FLUX。跨多种概念和多种擦除方法的实验表明，两种架构均能实现一致的概念再激活，凸显了CSR的跨架构适用性以及看似被擦除的概念的持续可恢复性。对于严格的裸体内容，CSR在FLUX上达到50.47%的平均攻击成功率（ASR），在Stable Diffusion上达到40.29%，在所有评估的安全设置中均排名第一。
英文摘要
Concept erasure aims to suppress undesirable knowledge in text-to-image generative models. However, existing robustness evaluations typically rely on relearning attacks tailored to specific model architectures. We study concept reactivation across two substantially different generative paradigms: noise-prediction U-Nets and flow-matching Transformers. We introduce \textbf{Concept Score Relearning (CSR)}, a unified parameter-level framework that reactivates erased concepts by optimizing each model within its native prediction space. CSR requires no external target-concept image dataset and applies the same concept-directed objective to both U-Net-based Stable Diffusion and Transformer-based FLUX. Experiments across diverse concepts and multiple erasure methods demonstrate consistent concept reactivation across both architectures, highlighting the cross-architecture applicability of CSR and the persistent recoverability of apparently erased concepts. For strict nudity, CSR reaches average ASRs of 50.47\% on FLUX and 40.29\% on Stable Diffusion, consistently ranking first across all evaluated safety settings.
