arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2602.00869cs.LGcs.AIcs.NAmath.NA
收藏

通过对齐流发散性来改进流匹配

Improving Flow Matching by Aligning Flow Divergence

Department of Mathematics, University of Utah, Salt Lake City, UT, USA(犹他大学数学系);Imaging (SCI) Institute, Salt Lake City, UT, USA(成像(SCI)研究所);Computer Science Division, 240 Argonne National Laboratory, Lemont, IL, USA(计算机科学部,阿贡国家实验室)

Yuhao Huang, Taos Transue, Shih-Hsin Wang, William Feldman, Hong Zhang, Bao Wang

AI总结:

本文提出了一种新的流匹配方法,通过同时匹配流及其发散性来提升生成模型的性能。

摘要:

条件流匹配(CFM)作为一种高效且无需模拟的方法,用于训练基于流的生成模型,实现了数据生成的显著性能。然而,CFM在学习概率路径的准确性上存在不足。在本文中,我们引入了一个新的偏微分方程来表征学习的概率路径与精确概率路径之间的误差,以及其解。我们表明,两个概率路径之间的总变分差距被CFM损失和一个相关发散损失的组合所限制。这一理论见解导致设计了一个新的目标函数,该函数同时匹配流及其发散性。我们的新方法通过显著提升基于流的生成模型的性能,而不牺牲生成效率。我们展示了这种增强的训练方法在多个重要基准任务上的优势,包括动态系统的生成建模、DNA序列和视频。代码可在https://github.com/Utah-Math-Data-Science/Flow_Div_Matching获取。

英文摘要:

Conditional flow matching (CFM) stands out as an efficient, simulation-free approach for training flow-based generative models, achieving remarkable performance for data generation. However, CFM is insufficient to ensure accuracy in learning probability paths. In this paper, we introduce a new partial differential equation characterization for the error between the learned and exact probability paths, along with its solution. We show that the total variation gap between the two probability paths is bounded above by a combination of the CFM loss and an associated divergence loss. This theoretical insight leads to the design of a new objective function that simultaneously matches the flow and its divergence. Our new approach improves the performance of the flow-based generative model by a noticeable margin without sacrificing generation efficiency. We showcase the advantages of this enhanced training approach over CFM on several important benchmark tasks, including generative modeling for dynamical systems, DNA sequences, and videos. Code is available at \href{https://github.com/Utah-Math-Data-Science/Flow_Div_Matching}{Utah-Math-Data-Science}.

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。

↑