遗忘的幻觉：通过初始潜在变量优化攻击未学习的扩散模型
The Illusion of Forgetting: Attack Unlearned Diffusion via Initial Latent Variable Optimization
AI总结：
本文提出IVO框架，通过优化初始潜在变量恢复未学习模型的噪声分布，揭示当前未学习方法的缺陷。
摘要：
文本到图像扩散模型（DMs）常被滥用以生成有害或受版权保护的内容，违反公众利益。概念擦除（去学习）是一种有前途的范式来缓解这一问题。然而，存在一种奇特的遗忘幻觉现象，其原因不明确。基于实证分析，我们正式解释了这一原因：大多数去学习部分破坏了语言符号与底层内部知识之间的映射，使知识保持为沉睡的记忆。我们进一步表明，在去噪过程中分布差异充当了衡量保留映射程度的可测量指标，也反映了去学习强度。受此启发，我们提出了IVO（初始潜在变量优化），一种新的攻击框架，旨在评估当前去学习方法的鲁棒性。IVO优化初始潜在变量以重新对齐去学习模型的噪声分布与其 vanilla 对应物的分布，从而重建破碎的映射，进而唤醒沉睡的记忆。广泛的实验涵盖了11种去学习技术和3种概念场景，显示IVO优于最先进的基线，揭示了当前去学习机制的基本缺陷。警告：本文包含可能让某些读者感到不适的图片。
英文摘要：
Text-to-image diffusion models (DMs) are frequently abused to produce harmful or copyrighted content, violating public interests. Concept erasure (unlearning) is a promising paradigm to alleviate this issue. However, there exists a peculiar forgetting illusion phenomenon with unclear cause. Based on empirical analysis, we formally explain this cause: most unlearning partially disrupt the mapping between linguistic symbols and the underlying internal knowledge, leaving the knowledge intact as dormant memories. We further demonstrate that distributional discrepancy in the denoising process serves as a measurable indicator of how much of the mapping is retained, also reflecting unlearning strength. Inspired by this, we propose IVO (Initial Latent Variable Optimization), a novel attack framework designed to assess the robustness of current unlearning methods. IVO optimizes initial latent variables to realign the noise distribution of unlearned models with that of their vanilla counterparts, which reconstructs the fractured mappings and consequently revives dormant memories. Extensive experiments covering 11 unlearning techniques and 3 concept scenarios show that IVO outperforms state-of-the-art baselines, exposing fundamental flaws in current unlearning mechanisms. Warning: This paper has unsafe images that may offend some readers.
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。
