研究扩散模型隐空间复用的极限，揭示何时可用旧隐空间，何时需重新学习。
On the Limits of Latent Reuse in Diffusion Models
- 分析隐空间复用时源与目标数据子空间的主角偏差影响
- 发现目标域得分误差由子空间错位和噪声放大共同决定
- 提出混合训练策略，指导共享隐空间维度选择
扩散模型常在低维隐空间中训练，并被复用于相关但分布偏移的数据集。本文研究在分布偏移下隐空间复用的可靠性。考虑源-目标设置，两者均近似低维但位于不同子空间。结果表明，冻结并复用源隐空间会引入目标域得分误差，该误差受两个因素制约：源与目标子空间间的主角偏差，以及扩散时间尺度放大的目标环境噪声。基于此限制，进一步研究混合源-目标训练，刻画所需共享隐空间维度如何依赖于两分布的相对几何结构。研究为隐空间复用的可靠性提供理论依据，并指明何时需学习共享表示。
原文摘要 · Abstract (English)
Diffusion models are often trained in low-dimensional latent spaces, which are then reused for related but shifted datasets. In this work, we study when such latent reuse remains reliable under distribution shift. We consider a source-target setting in which both datasets are approximately low-dimensional but may lie near different subspaces. We show that freezing and reusing a source latent space induces a target-domain score error governed by two quantities: the principal-angle misalignment between the source and target subspaces, and the target ambient noise amplified by the diffusion time scale. Motivated by these limits, we further study mixed source-target training and characterize how the required shared latent dimension depends on the relative geometry of the two distributions. Our results provide theoretical guidance on when latent reuse is reliable and when learning a shared representation may be necessary.
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。