发表机构
Institute of Science Tokyo; Tohoku University; Kumamoto University; Sigma-i Co., Ltd.（东京科学大学; 东北大学; 熊本大学; Sigma-i股份有限公司）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文提出一种参考过程设计框架，可同时实现无模拟训练与有限时间生成，还揭示得分匹配并非扩散模型训练的基础，且条件流匹配是该框架的小噪声极限。
AI 中文摘要
生成式扩散模型的性能由连接经验分布与先验分布的参考扩散过程的选择决定。传统方法通常需在无模拟训练与有限时间生成之间做权衡。我们提出一种参考过程设计框架，可同时实现这两种目标。核心思路是指定易处理的时变条件分布，再构建以这些分布为边缘分布的参考过程。该框架表明，得分匹配并非扩散模型训练的基础，而是通过参考过程的逆转自然产生的。我们进一步证明，条件流匹配是所提框架的小噪声极限。
英文摘要
The performance of generative diffusion models is determined by the choice of the reference diffusion process connecting the empirical and prior distributions. Conventional approaches typically trade off simulation-free training against finite-time generation. We propose a framework for designing the reference process that achieves both simultaneously. The key idea is to prescribe tractable time-dependent conditional distributions and then construct the reference process realizing them as its marginals. This framework reveals that score matching is not fundamental to diffusion-model training but instead emerges naturally through reversal of the reference process. We further show that conditional flow matching arises as the small-noise limit of the proposed framework.
