Markovian Flow Matching: Accelerating MCMC with Continuous Normalizing Flows

arXiv:2405.14392v1 Announce Type: cross
Abstract: Continuous normalizing flows (CNFs) learn the probability path between a reference and a target density by modeling the vector field generating said path using neural networks. Recently, Lipman et al. (2022) introduced a simple and inexpensive method for training CNFs in generative modeling, termed flow matching (FM). In this paper, we re-purpose this method for probabilistic inference by incorporating Markovian sampling methods in evaluating the FM objective and using the learned probability path to improve …

abstract arxiv continuous continuous normalizing flows cs.lg flow generative generative modeling learn mcmc modeling networks neural networks path probability reference said simple stat.me stat.ml training type vector