Riemannian流匹配(RFM)通过学习一个时间依赖的切向量场,将基于流的生成模型扩展到支撑在流形上的数据。我们开发了一种非渐近的总变分(TV)收敛分析,用于使用学习的向量场和欧拉离散化在流形上进行采样。我们的关键技术成分是支配两个流形ODE流之间TV演化的微分不等式,它通过向量场不匹配的散度和参考流的分数来表达TV的时间导数;控制这些项需要建立新的界限,这些界限明确考虑了平行运输和曲率。在对流匹配场的总体光滑性假设以及对学习场的均匀(紧致流形)或均方(Hadamard流形)近似保证下,我们得到形式为TV ≤ C_Lip h + C_ε ε(在紧致流形上还有一个额外的更高阶ε²项)的显式界限,清晰地分离了数值离散化和学习误差。实例化结果在超球面S^d上和满足温和矩条件的SPD(n)流形上得到显式的多项式迭代复杂度。
英文摘要:
Riemannian flow matching (RFM) extends flow-based generative modeling to data supported on manifolds by learning a time-dependent tangent vector field whose flow-ODE transports a simple base distribution to the data law. We develop a nonasymptotic Total Variation (TV) convergence analysis for RFM samplers that use a learned vector field together with Euler discretization on manifolds. Our key technical ingredient is a differential inequality governing the evolution of TV between two manifold ODE flows, which expresses the time-derivative of TV through the divergence of the vector-field mismatch and the score of the reference flow; controlling these terms requires establishing new bounds that explicitly account for parallel transport and curvature. Under smoothness assumptions on the population flow-matching field and either uniform (compact manifolds) or mean-square (Hadamard manifolds) approximation guarantees for the learned field, we obtain explicit bounds of the form $\mathrm{TV}\le C_{\mathrm{Lip}}\,h + C_{\varepsilon}\,\varepsilon$ (with an additional higher-order $\varepsilon^2$ term on compact manifolds), cleanly separating numerical discretization and learning errors. Here, $h$ is the step-size and $\varepsilon$ is the target accuracy. Instantiations yield \emph{explicit} polynomial iteration complexities on the hypersphere $S^d$, and on the SPD$(n)$ manifolds under mild moment conditions.
Riemannian Flow Matching Policy for Robot Motion Learning
Institute for Anthropomatics and Robotics, Karlsruhe Institute of Technology(卡尔斯鲁厄理工学院人形机器人与智能系统研究所);Bosch Center for Artificial Intelligence(博世人工智能中心)
Fast and Robust Visuomotor Riemannian Flow Matching Policy
Bosch Center for Artificial Intelligence(博世人工智能中心);Division of Robotics, Perception, and Learning, KTH Royal Institute of Technology(机器人、感知与学习 division,皇家理工学院);Computer Science Department of the Technische Universität Darmstadt(达姆施塔特技术大学计算机科学系)
DisRFM: Polar Riemannian Flow Matching for Structure-Preserving Graph Domain Adaptation
The Chinese University of Hong Kong(香港中文大学);National University of Defense Technology(国防科学技术大学);Hebei University of Technology(河北工业大学);City University of Hong Kong(香港城市大学)
Matcha: Multi-Stage Riemannian Flow Matching for Accurate and Physically Valid Molecular Docking
Ligand Pro;Skolkovo Institute of Science and Technology(斯克洛洛沃科学与技术研究院);Artificial Intelligence Center(人工智能中心);Center for Molecular and Cellular Biology(分子与细胞生物学中心);Artificial Intelligence Research Institute(人工智能研究 institute)
Riemannian Motion Generation: A Unified Framework for Human Motion Representation and Generation via Riemannian Flow Matching
Department of Data Science and Artificial Intelligence, The Hong Kong Polytechnic University(数据科学与人工智能系,香港理工大学);Department of Applied Mathematics, The Hong Kong Polytechnic University(应用数学系,香港理工大学);Department of Data Science and Statistics, Southern University of Science and Technology(数据科学与统计系,南方科技大学)