提出直接学习全局转移流，实现单步生成和任意时间点采样。
Transition Flow Matching
- 直接建模全局转移流，替代传统局部速度场学习
- 支持单步生成，且可在任意时间点进行采样
- 与均值速度流统一理论框架，兼具理论深度与实用性
主流流匹配方法通常聚焦于学习局部速度场，这在生成过程中必然需要多步积分。相比之下，均值速度流通过数学严谨的公式建立局部速度场与全局均值速度之间的关系，使后者可被学习，并允许生成过程转移到任意未来时间点。本文提出一种新范式：直接学习转移流。作为全局量，转移流天然支持单步生成或在任意时间点进行采样。此外，我们揭示了该方法与均值速度流之间的联系，建立了统一的理论视角。大量实验验证了方法的有效性，并支持我们的理论主张。
原文摘要 · Abstract (English)
Mainstream flow matching methods typically focus on learning the local velocity field, which inherently requires multiple integration steps during generation. In contrast, Mean Velocity Flow models establish a relationship between the local velocity field and the global mean velocity, enabling the latter to be learned through a mathematically grounded formulation and allowing generation to be transferred to arbitrary future time points. In this work, we propose a new paradigm that directly learns the transition flow. As a global quantity, the transition flow naturally supports generation in a single step or at arbitrary time points. Furthermore, we demonstrate the connection between our approach and Mean Velocity Flow, establishing a unified theoretical perspective. Extensive experiments validate the effectiveness of our method and support our theoretical claims.
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。