发现条件扩散模型实际生成偏离理想去噪，影响采样算法效果。
Is Your Conditional Diffusion Model Actually Denoising?
- 提出「调度偏差」度量标准，量化模型偏离去噪过程的程度。
- 无论模型大小或训练数据多少，偏差现象均持续存在。
- 揭示平滑性归纳偏置导致不同条件区间去噪流难以衔接。
我们研究了带条件变量的扩散模型的归纳偏置，这类模型广泛应用于文本条件生成图像和观测条件下的连续控制策略。发现当这些模型进行条件查询时，其生成结果始终偏离扩散模型理论基础中的理想去噪过程，导致主流采样算法（如DDPM、DDIM）之间产生分歧。本文提出「调度偏差」这一严格度量，用于捕捉偏离标准去噪过程的速率，并提供计算方法。关键发现是，这种偏离现象与模型容量或训练数据量无关。我们推测该现象源于在条件空间不同区域间桥接不同去噪流的困难，并从理论上证明了平滑性归纳偏置如何导致此类偏差产生。
原文摘要 · Abstract (English)
We study the inductive biases of diffusion models with a conditioning-variable, which have seen widespread application as both text-conditioned generative image models and observation-conditioned continuous control policies. We observe that when these models are queried conditionally, their generations consistently deviate from the idealized "denoising" process upon which diffusion models are formulated, inducing disagreement between popular sampling algorithms (e.g. DDPM, DDIM). We introduce Schedule Deviation, a rigorous measure which captures the rate of deviation from a standard denoising process, and provide a methodology to compute it. Crucially, we demonstrate that the deviation from an idealized denoising process occurs irrespective of the model capacity or amount of training data. We posit that this phenomenon occurs due to the difficulty of bridging distinct denoising flows across different parts of the conditioning space and show theoretically how such a phenomenon can arise through an inductive bias towards smoothness.
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。