mlsanityData to 14 Oct 2026

Federated Flow Matching

arXiv 2509.21250 ↗First posted 25 Sep 2025cs.LGcited by 0

On the map: Diffusion models → Robustness, privacy and adversarial aspects

Abstract (arXiv)

Data today is decentralized, generated and stored across devices and institutions where privacy, ownership, and regulation prevent centralization. This motivates the need to train generative models directly from distributed data locally without central aggregation. In this paper, we introduce Federated Flow Matching (FFM), a framework for training flow matching models under privacy constraints. Specifically, we first examine FFM-vanilla, where each client trains locally with independent source and target couplings, preserving privacy but yielding curved flows that slow inference. We then develop FFM-LOT, which employs local optimal transport couplings to improve straightness within each client but lacks global consistency under heterogeneous data. Finally, we propose FFM-GOT, a federated strategy based on the semi-dual formulation of optimal transport, where a shared global potential function coordinates couplings across clients. Experiments on synthetic and image datasets show that FFM enables privacy-preserving training while enhancing both the flow straightness and sample quality in federated settings, with performance comparable to the centralized baseline.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Robustness, privacy and adversarial aspects, one of 14 directions in Diffusion models. Builds on 24 of its 37 references that are in the map. First posted on arXiv 25 Sep 2025.

Builds on

24 of 37 references.

Closest ancestors · 3

Same direction, newest first

  1. 2025Federated Learning for Diffusion Models9 Mar
  2. 2024FedDM: Enhancing Communication Efficiency and Handling Data Heterogeneity in Federated Diffusion Models20 Jul
  3. 2024Training Diffusion Models with Federated Learning18 Jun

Other references · 21

  1. 2020Denoising Diffusion Probabilistic Modelscited by 21,076
  2. 2020Score-Based Generative Modeling through Stochastic Differential Equationscited by 9,122
  3. 2016Communication-Efficient Learning of Deep Networks from Decentralized Datacited by 7,669
  4. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,058
  5. 2022Flow Matching for Generative Modelingcited by 5,702
  6. 2022Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flowcited by 3,700
  7. 2016Federated Learning: Strategies for Improving Communication Efficiencycited by 1,475
  8. 2023Improving and generalizing flow-based generative models with minibatch optimal transportcited by 879
  9. 2022Rectified Flow: A Marginal Preserving Approach to Optimal Transportcited by 283
  10. 2017Joint Distribution Optimal Transportation for Domain Adaptationcited by 282
  11. 2016Stochastic Optimization for Large-scale Optimal Transportcited by 189
  12. 2020TrajectoryNet: A Dynamic Optimal Transport Network for Modeling Cellular Dynamicscited by 160
9 more
  1. 2023Diffusion Schr"odinger Bridge Matchingcited by 153
  2. 2018Improving GANs Using Optimal Transportcited by 123
  3. 2019Generative Models for Effective ML on Private, Decentralized Datasetscited by 85
  4. 2024Optimal Flow Matching: Learning Straight Trajectories in Just One Stepcited by 78
  5. 2023Generative Modeling through the Semi-dual Formulation of Unbalanced Optimal Transportcited by 41
  6. 2023GENOT: Entropic (Gromov) Wasserstein Flow Matching with Applications to Single-Cell Genomicscited by 28
  7. 2019Quantitative stability of optimal transport maps and linearization of the 2-Wasserstein spacecited by 25
  8. 2025Joint Velocity-Growth Flow Matching for Single-Cell Dynamics Modelingcited by 20
  9. 2025On Fitting Flow Models with Large Sinkhorn Couplingscited by 11

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026FedFG: Privacy-Preserving and Robust Federated Learning via Flow-Matching Generation
  2. 2026Deep Leakage with Generative Flow Matching Denoiser
  3. 2024Federated Learning Nodes Can Reconstruct Peers' Image Data
  4. 2025Privacy-Preserving Decentralized Federated Learning via Explainable Adaptive Differential Privacy
  5. 2025FedRand: Enhancing Privacy in Federated Learning with Randomized LoRA Subparameter Updates
  6. 2026ERIS: Enhancing the Privacy-Utility-Scalability Trade-Off in Federated Learning via Federated Shard Aggregation
  7. 2025Parameter-Efficient and Personalized Federated Training of Generative Models at the Edge
  8. 2025Federated Learning: A Survey on Privacy-Preserving Collaborative Intelligence