Neural flow matching models

Denoising diffusion probabilistic models (DDPMs), score-based generative models, generative diffusion processes, neural energy models…

2021-11-10 — 2025-05-28

approximation
Bayes
generative
Monte Carlo
neural nets
optimization
probabilistic algorithms
probability
score function
statistics

Suspiciously similar content

A close cousin to neural denoising diffusion models.

Figure 1

1 Tutorials of note

  • Scott Hawley, Flow With What You Know

  • Lipman et al. (2024)

    Flow Matching (FM) is a recent framework for generative modeling that has achieved state-of-the-art performance across various domains, including image, video, audio, speech, and biological structures. This guide offers a comprehensive and self-contained review of FM, covering its mathematical foundations, design choices, and extensions. By also providing a PyTorch package featuring relevant examples (e.g., image and text generation), this work aims to serve as a resource for both novice and experienced researchers interested in understanding, applying and further developing FM

    facebookresearch/flow_matching.

2 Conditioning

Keyword: dependent coupling.

3 Discrete state

Start from Eijkelboom et al. (2024) ?

4 References

Cheng, Han, Maddix, et al. 2024. “Hard Constraint Guided Flow Matching for Gradient-Free Generation of PDE Solutions.”
Dax, Wildberger, Buchholz, et al. 2023. “Flow Matching for Scalable Simulation-Based Inference.”
Eijkelboom, Bartosh, Naesseth, et al. 2024. “Variational Flow Matching for Graph Generation.”
Kerrigan, Migliorini, and Smyth. 2024. “Functional Flow Matching.” In Proceedings of The 27th International Conference on Artificial Intelligence and Statistics.
Köhler, Chen, Krämer, et al. 2023. “Flow-Matching: Efficient Coarse-Graining of Molecular Dynamics Without Forces.” Journal of Chemical Theory and Computation.
Lipman, Chen, Ben-Hamu, et al. 2023. “Flow Matching for Generative Modeling.” In.
Lipman, Havasi, Holderrieth, et al. 2024. “Flow Matching Guide and Code.”