DyME：基于双层正交LoRA适配的扩散模型动态多概念擦除
DyME: Dynamic Multi-Concept Erasure in Diffusion Models with Bi-Level Orthogonal LoRA Adaptation
AI总结：
针对扩散模型静态多概念擦除的局限，提出DyME框架，通过训练概念特定LoRA适配器并按需动态组合，结合双层正交约束减少干扰，在ErasureBench-H等基准上实现更高保真度的多概念擦除。
摘要：
文本到图像扩散模型（DMs）会无意中复制受版权保护的风格和受保护的视觉概念，引发法律和伦理问题。概念擦除作为一种保护措施应运而生，旨在通过微调选择性地抑制此类概念。然而，现有方法无法扩展到实际场景，在这些场景中，提供商必须擦除多个且可能相互冲突的概念。核心瓶颈在于它们依赖静态擦除：即使用单个检查点进行微调以移除所有目标概念，而不考虑推理时的实际擦除需求。这种刚性设计与现实使用不匹配，因为每次生成的请求各不相同，导致擦除成功率下降，并降低非目标内容的保真度。我们提出DyME，一种按需擦除框架，它训练轻量级的、特定于概念的LoRA适配器，并在推理时仅动态组合所需适配器。这种模块化设计实现了灵活的多概念擦除，但朴素组合会导致适配器间干扰，尤其是在抑制多个或语义相关的概念时。为克服这一问题，我们在特征和参数层面引入双层正交性约束，解耦表示偏移并强制正交适配器子空间。我们进一步开发了ErasureBench-H，一个具有品牌-系列-角色结构的新型分层基准，支持跨语义粒度和擦除集大小的原则性评估。在ErasureBench-H和标准数据集（如CIFAR-100、Imagenette）上的实验表明，DyME持续优于最先进的基线，以最小的附带退化实现更高的多概念擦除保真度。
英文摘要：
Text-to-image diffusion models (DMs) inadvertently reproduce copyrighted styles and protected visual concepts, raising legal and ethical concerns. Concept erasure has emerged as a safeguard, aiming to selectively suppress such concepts through fine-tuning. However, existing methods do not scale to practical settings where providers must erase multiple and possibly conflicting concepts. The core bottleneck is their reliance on static erasure: a single checkpoint is fine-tuned to remove all target concepts, regardless of the actual erasure needs at inference. This rigid design mismatches real-world usage, where requests vary per generation, leading to degraded erasure success and reduced fidelity for non-target content. We propose DyME, an on-demand erasure framework that trains lightweight, concept-specific LoRA adapters and dynamically composes only those needed at inference. This modular design enables flexible multi-concept erasure, but naive composition causes interference among adapters, especially when many or semantically related concepts are suppressed. To overcome this, we introduce bi-level orthogonality constraints at both the feature and parameter levels, disentangling representation shifts and enforcing orthogonal adapter subspaces. We further develop ErasureBench-H, a new hierarchical benchmark with brand-series-character structure, enabling principled evaluation across semantic granularities and erasure set sizes. Experiments on ErasureBench-H and standard datasets (e.g., CIFAR-100, Imagenette) demonstrate that DyME consistently outperforms state-of-the-art baselines, achieving higher multi-concept erasure fidelity with minimal collateral degradation.
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。
