Research map / Image generation / Fast diffusion sampling
Gauge Flow Matching: Efficient Constrained Generative Modeling over General Convex Set and Beyond
Xinpeng Li, Enming Liang, Minghua Chen
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add gauge mapping to flow matching (GFM instead of FM) | flow matching · convex sets feasibility | feasibility (%): 95.9 -> 100 | Improved |
| use gauge mapping instead of regular reflection | constrained flow matching · convex sets | inference time (s): 14.0 -> 0.63 | Improved |
| use gauge mapping instead of projection | constrained flow matching · star-convex sets | inference time (s): 347 -> 0.74 | Improved |
| use gauge mapping instead of Metropolis sampling | constrained flow matching · convex sets | MMD (x10^-3): 130 -> 3.50 | Improved |
| use gauge mapping instead of Metropolis | constrained flow matching · geodesic-convex sets | MMD (x10^-3): 48.3 -> 1.29 | Improved |
| add gauge mapping to FM on geodesic-convex sets | flow matching · geodesic-convex sets | MMD (x10^-3): 1.20 -> 1.29 | No clear effect |
| use gauge mapping instead of projection on time series | constrained flow matching · PEMS-BAY | inference time (s): 49.7 -> 0.43 | Improved |
| add gauge mapping to FM on time series | flow matching · PEMS-BAY | feasibility (%): 88.5 -> 100 | Improved |
| use gauge mapping instead of projection for watermark | constrained flow matching · CIFAR-10 watermark | FID: 6.88 -> 5.85 | Improved |
| add constraints via gauge mapping to FM for watermark | flow matching · CIFAR-10 watermark | FID: 3.57 -> 5.85 | Hurt |
| add gauge mapping to FM on 10x10 combinatorial problems | flow matching · relaxed combinatorial n = 10 x 10 | feasibility (%): 0 -> 100 | Improved |
| choose central interior point instead of near-boundary (ablation) | GFM · synthetic generation | MMD | Improved |
Closest papers by meaning
- Physics-Constrained Flow Matching: Sampling Generative Models with Hard Constraints (NeurIPS 2025 · similarity 0.88)
- Beckmann Transport Models: From Autonomous Flows to One-Step Maps (NeurIPS 2026 · similarity 0.88)
- Flow Matching on General Geometries (ICLR 2024 · similarity 0.87)
- Gradient-Free Generation for Hard-Constrained Systems (ICLR 2025 · similarity 0.87)
- Optimal Flow Matching: Learning Straight Trajectories in Just One Step (NeurIPS 2024 · similarity 0.87)
- On the Guidance of Flow Matching (ICML 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
