Abstract:Flow matching (FM) is a family of training algorithms for fitting continuous normalizing flows (CNFs). A standard approach to FM, called conditional flow matching (CFM), exploits the fact that the marginal vector field of a CNF can be learned by fitting least-square regression to the so-called conditional vector field specified given one or both ends of the flow path. We show that viewing CFM training from a Bayesian decision theoretic perspective on parameter estimation opens the door to generalizations of CFM algorithms. We propose one such extension by introducing a CFM algorithm based on defining conditional probability paths given what we refer to as ``streams'', instances of latent stochastic paths that connect pairs of noise and observed data. Further, we advocate the modeling of these latent streams using Gaussian processes (GPs). The unique distributional properties of GPs, and in particular the fact that the velocity of a GP is still a GP, allows drawing samples from the resulting stream-augmented conditional probability path without simulating the actual streams, and hence the ``simulation-free" nature of CFM training is preserved. We show that this generalization of the CFM can substantially reduce the variance in the estimated marginal vector field at a moderate computational cost, thereby improving the quality of the generated samples under common metrics. Additionally, we show that adopting the GP on the streams allows for flexibly linking multiple related training data points (e.g., time series) and incorporating additional prior information. We empirically validate our claim through both simulations and applications to two hand-written image datasets.

Conditional Wasserstein Distances with Applications in Bayesian OT Flow Matching

Y-Diagonal Couplings: Approximating Posteriors with Conditional Wasserstein Distances

Dynamic Conditional Optimal Transport through Simulation-Free Flows

Wasserstein Flow Matching: Generative modeling over families of distributions

Conditional Variable Flow Matching: Transforming Conditional Densities with Amortized Conditional Optimal Transport

Flow Matching in Latent Space

Metric Flow Matching for Smooth Interpolations on the Data Manifold

Stream-level flow matching from a Bayesian decision theoretic perspective

Bayesian Inverse Problems with Conditional Sinkhorn Generative Adversarial Networks in Least Volume Latent Spaces

Flow Priors for Linear Inverse Problems via Iterative Corrupted Trajectory Matching

Hessian-Informed Flow Matching

Wasserstein Generative Learning of Conditional Distribution

Sliced-Wasserstein Distances and Flows on Cartan-Hadamard Manifolds

Deep conditional distribution learning via conditional Föllmer flow

Convergence Analysis for General Probability Flow ODEs of Diffusion Models in Wasserstein Distances

Nonparametric Generative Modeling with Conditional Sliced-Wasserstein Flows

Convergence analysis of OT-Flow for sample generation

Minibatch Optimal Transport and Perplexity Bound Estimation in Discrete Flow Matching

ProbFlow: Joint Optical Flow and Uncertainty Estimation

Flow Matching with Gaussian Process Priors for Probabilistic Time Series Forecasting