让图像生成模型精准删除目标概念，同时保护附近相关概念的细节。
Neighbor-Aware Localized Concept Erasure in Text-to-Image Diffusion Models
- 通过谱加权嵌入调制与注意力引导门控，分三步精准定位并清除目标概念
- 在细粒度数据集上删除目标概念后，相邻类别保留率提升23%以上
- 无需训练即可通用，适合需精确控制生成内容的场景
文本到图像扩散模型中的概念擦除旨在移除不希望出现的概念，同时保持整体生成能力。局部擦除方法试图将修改限制在目标概念占据的空间区域。然而我们发现，抑制一个概念可能无意中削弱语义相关的邻近概念，导致细粒度领域中的生成质量下降。为此，我们提出无需训练的邻居感知局部概念擦除（NLCE）框架，以更好地在移除目标概念的同时保护邻近概念。该方法包含三个阶段：(1) 谱加权嵌入调制，减弱目标概念方向并稳定邻近概念表示；(2) 注意力引导的空间门控，识别仍存在概念激活的区域；(3) 空间门控的硬擦除，仅在必要位置消除残留痕迹。这一邻居感知流程可在保持周围概念结构的前提下实现局部概念删除。在细粒度数据集（Oxford Flowers、Stanford Dogs）上的实验表明，该方法能有效移除目标概念，同时更好保留紧密相关的类别。在名人身份、敏感内容和艺术风格等场景的额外结果验证了其鲁棒性与泛化能力。
原文摘要 · Abstract (English)
Concept erasure in text-to-image diffusion models seeks to remove undesired concepts while preserving overall generative capability. Localized erasure methods aim to restrict edits to the spatial region occupied by the target concept. However, we observe that suppressing a concept can unintentionally weaken semantically related neighbor concepts, reducing fidelity in fine-grained domains. We propose Neighbor-Aware Localized Concept Erasure (NLCE), a training-free framework designed to better preserve neighboring concepts while removing target concepts. It operates in three stages: (1) a spectrally-weighted embedding modulation that attenuates target concept directions while stabilizing neighbor concept representations, (2) an attention-guided spatial gate that identifies regions exhibiting residual concept activation, and (3) a spatially-gated hard erasure that eliminates remaining traces only where necessary. This neighbor-aware pipeline enables localized concept removal while maintaining the surrounding concept neighborhood structure. Experiments on fine-grained datasets (Oxford Flowers, Stanford Dogs) show that our method effectively removes target concepts while better preserving closely related categories. Additional results on celebrity identity, explicit content and artistic style demonstrate robustness and generalization to broader erasure scenarios.
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。