擦除或侵蚀？评估未学习文本到图像扩散模型中的组合退化
Erasure or Erosion? Evaluating Compositional Degradation in Unlearned Text-To-Image Diffusion Models
AI总结：
本文通过组合文本到图像生成视角，评估未学习方法对 nudity 的影响，发现去除非目标概念与组合完整性之间存在权衡，现有评估方法存在局限。
摘要：
后验未学习已成为从大规模文本到图像扩散模型中移除不受欢迎概念的实用机制。然而，先前工作主要通过擦除成功率评估未学习，其对更广泛生成能力的影响尚不明确。本文通过系统实证研究，从组合文本到图像生成的角度评估概念未学习。聚焦于Stable Diffusion 1.4中的 nudity 移除，我们使用T2I-CompBench++和GenEval以及已建立的未学习基准，评估了一组最先进的未学习方法。我们的结果揭示了未学习有效性与组合完整性之间的持续权衡：实现强擦除的方法通常会带来属性绑定、空间推理和计数方面的显著退化。相反，保持组合结构的方法往往无法提供稳健的擦除。这些发现突显了当前评估实践的局限，并强调了需要未学习目标，这些目标应明确考虑语义保持，而不仅仅是目标抑制。
英文摘要：
Post-hoc unlearning has emerged as a practical mechanism for removing undesirable concepts from large text-to-image diffusion models. However, prior work primarily evaluates unlearning through erasure success; its impact on broader generative capabilities remains poorly understood. In this work, we conduct a systematic empirical study of concept unlearning through the lens of compositional text-to-image generation. Focusing on nudity removal in Stable Diffusion 1.4, we evaluate a diverse set of state-of-the-art unlearning methods using T2I-CompBench++ and GenEval, alongside established unlearning benchmarks. Our results reveal a consistent trade-off between unlearning effectiveness and compositional integrity: methods that achieve strong erasure frequently incur substantial degradation in attribute binding, spatial reasoning, and counting. Conversely, approaches that preserve compositional structure often fail to provide robust erasure. These findings highlight limitations of current evaluation practices and underscore the need for unlearning objectives that explicitly account for semantic preservation beyond targeted suppression.
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。
