Research map / Probabilistic and scientific modeling / Variational and generative sampling
Parsimonious Flow Matching for Efficient Image Generation
Tianjiao Ding, Ziqing Xu, Benjamin Haeffele, Hongkang Li, Rene Vidal
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 |
|---|---|---|---|
| replace isotropic Gaussian latent with learned MoG latent | OT-CFM · CIFAR-10 (NFE=4) | FID: 9.20 -> 8.52 | Improved |
| replace isotropic Gaussian latent with learned MoG latent | OT-CFM · CIFAR-10 (NFE=12) | FID: 6.24 -> 5.71 | Improved |
| use MoG latent (PFM) vs MFM-BatchOT | flow matching · ImageNet 32x32 (NFE=4) | FID: 17.28 -> 9.90 | Improved |
| use MoG latent | OT-CFM · CIFAR-10 training efficiency | iterations to equal FID: 100K -> 50K iterations (NFE=2) | Improved |
| fit MoG with k-means (10 modes) instead of random MoG (ablation) | PFM · CIFAR-10 (NFE=12, 100k) | FID: 5.85 -> 5.71 | Improved |
| fit MoG with class labels (supervised) instead of k-means (ablation) | PFM · CIFAR-10 (NFE=12, 100k) | FID: 5.71 -> 6.3 | Hurt |
| add eigenvalue thresholding of per-mode covariances (ablation) | PFM · CIFAR-10 | FID | Improved |
| class-conditional OT instead of global OT coupling (ablation) | PFM · CIFAR-10 | FID | No clear effect |
Closest papers by meaning
- Gaussian Mixture Flow Matching Models (ICML 2025 · similarity 0.90)
- Blockwise Flow Matching: Improving Flow Matching Models For Efficient High-Quality Generation (NeurIPS 2025 · similarity 0.89)
- Structured Coupling for Flow Matching (NeurIPS 2026 · similarity 0.89)
- Value Gradient Guidance for Flow Matching Alignment (NeurIPS 2025 · similarity 0.88)
- Stream-level Flow Matching with Gaussian Processes (ICML 2025 · similarity 0.88)
- Coreset-Induced Conditional Velocity Flow Matching (NeurIPS 2026 · similarity 0.88)
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.
