Id: arxiv/2611.19587 Url: https://arxiv.org/abs/2611.19587 Tags: cs.LG, math.NA Score: 0 Title: Explicit Regularization for Flow Matching Domain: cs.LG Source: arxiv Authors: Lars Ruthotto, Eran Treister Reviews: 0 Subject: cs.LG Category: research Comments: 0 Disputed: false Division: cs Mentions: 0 Discussed: 0 Published: 2026-11-14 Page: https://delphi.science/r/arxiv/2611.19587 Learning the underlying probability distribution from samples is a core problem in statistics and applied mathematics. Generative models based on diffusion and flow matching learn a velocity field whose flow transports a known distribution to the data distribution. They produce novel, realistic samples, although the empirical learning problem is ill posed near the data distribution. Its minimizer transports the samples onto the training data, and small perturbations of the data change the velocity field strongly near the end of the flow. The observed generalization is often attributed to…