← 研踪 PaperTrailarXiv:2611.26407 · v1
Flow Matching from Non-Gaussian Sources: A Theory of Efficient Sampling
首次提交 2026-11-20 00:54
版本更新 2026-11-20 00:54
雷达观测 2026-11-20 12:52
摘要
Flow matching is a versatile framework for generative modeling that transports a source distribution to a target data distribution. A key advantage is its flexibility in the choice of source distribution, such as heavy-tailed and multimodal sources. Yet finite-step sampling guarantees for non-Gaussian sources remain limited. We establish a nonasymptotic convergence theory for Euler discretization of flow matching with linear interpolation. Suppose the source distribution on has an everywhere positive, continuously differentiable density, finite total variance , and an -Lipschitz score. Using the population velocity field and a suitable time grid, this sampler achieves total variation error at most in iterations, provided the target distribution satisfies a mild second-moment condition. These assumptions cover, among many examples, source distributions such as multivariate Student's distributions with more than two degrees of freedom and finite Gaussian mixtures with shared covariance, while accommodating target distributions with neither a smooth density nor bounded support.
作者备注:36 pages
观测到的版本
相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。
v1 · 2026-11-20 00:54 · 最新观测版本
Flow Matching from Non-Gaussian Sources: A Theory of Efficient Sampling
Flow matching is a versatile framework for generative modeling that transports a source distribution to a target data distribution. A key advantage is its flexibility in the choice of source distribution, such as heavy-tailed and multimodal sources. Yet finite-step sampling guarantees for non-Gaussian sources remain limited. We establish a nonasymptotic convergence theory for Euler discretization of flow matching with linear interpolation. Suppose the source distribution on has an everywhere positive, continuously differentiable density, finite total variance , and an -Lipschitz score. Using the population velocity field and a suitable time grid, this sampler achieves total variation error at most in iterations, provided the target distribution satisfies a mild second-moment condition. These assumptions cover, among many examples, source distributions such as multivariate Student's distributions with more than two degrees of freedom and finite Gaussian mixtures with shared covariance, while accommodating target distributions with neither a smooth density nor bounded support.
查看此版本解读 →查看此版本原文 ↗