扩散模型中原型引导的概念擦除
Prototype-Guided Concept Erasure in Diffusion Models
利用扩散模型文本嵌入空间的几何结构，通过聚类得到概念原型，并在推理时作为负条件信号，为概念擦除和负向引导提供了一种无需微调的通用手段，可直接迁移到其他需要抑制特定语义的生成控制场景。
原文依据 we exploit the model's intrinsic embedding geometry to identify latent embeddings that encode a given concept
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论概念擦除被广泛用于图像生成，以防止文本到图像模型生成不期望的内容。现有方法能够有效擦除具体而狭窄的概念，例如特定的知识产权（如皮卡丘）或可识别的角色（如埃隆·马斯克）。然而，对于“性”或“暴力”这类宽泛概念，其性能会下降，因为这些概念范围广泛且具有多面性，难以可靠地擦除。为克服这一局限，我们利用模型内在的嵌入几何结构来识别编码给定概念的潜变量嵌入。通过对这些嵌入进行聚类，我们得到一组概念原型，用以概括模型对该概念的内部表示，并在推理时将其作为负条件信号，从而实现精确且可靠的擦除。在多个基准上的大量实验表明，我们的方法在更可靠地移除宽泛概念的同时保持了整体图像质量，朝着更安全、更可控的图像生成迈出了一步。
英文原文摘要 1108 字符
Concept erasure is extensively utilized in image generation to prevent text-to-image models from generating undesired content. Existing methods can effectively erase narrow concepts that are specific and concrete, such as distinct intellectual properties (e.g. Pikachu) or recognizable characters (e.g. Elon Musk). However, their performance degrades on broad concepts such as "sexual" or "violent", whose wide scope and multi-faceted nature make them difficult to erase reliably.To overcome this limitation, we exploit the model's intrinsic embedding geometry to identify latent embeddings that encode a given concept. By clustering these embeddings, we derive a set of concept prototypes that summarize the model's internal representations of the concept, and employ them as negative conditioning signals during inference to achieve precise and reliable erasure. Extensive experiments across multiple benchmarks show that our approach achieves substantially more reliable removal of broad concepts while preserving overall image quality, marking a step towards safer and more controllable image generation.