Research map / Image generation / Fast diffusion sampling
Diffusion and Flow-based Copulas: Forgetting and Remembering Dependencies
David Huk, Theodoros Damoulas
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 |
|---|---|---|---|
| use classification-diffusion copula instead of Ratio copula for density | copula density estimation · Magic | log-likelihood: 6.76 -> 18.65 | Improved |
| use classification-diffusion copula instead of Ratio copula | copula density estimation · Dry_Bean | log-likelihood: 48.21 -> 50.21 | Improved |
| use classification-diffusion copula instead of Ratio copula | copula density estimation · MNIST | log-likelihood: 334.42 -> 346.70 | Improved |
| use reflection or c_dc copula instead of Gaussian/vine for samples | copula sampling · Robocup | Wasserstein-2: 3.96 -> 3.84 | Improved |
| use proposed copulas instead of Gaussian copula for Kendall-tau fit | dependence structure · Robocup | Frobenius norm: 0.45 -> 0.49 (reflection), 0.51 (c_dc) | Hurt |
| use c_dc instead of Gaussian copula | sample quality · MNIST | FID: 102.56 -> 7.38 | Improved |
| use reflection copula instead of best baseline | sample quality · Cifar cars/trucks | FID: 100.04 -> 42.14 | Improved |
| use c_dc instead of Gaussian copula on digits FID | sample quality · digits | FID: 5.74 -> 15.24 | Hurt |
| use reflection copula instead of Gaussian on Cifar W2 | sample quality · Cifar cars/trucks | Wasserstein-2: 30.62 -> 32.40 | Hurt |
| use vine copula for density | Cifar density · Cifar cars/trucks | log-likelihood: NaN | Hurt |
Closest papers by meaning
- Discrete Copula Diffusion (ICLR 2025 · similarity 0.89)
- Probing the Latent Hierarchical Structure of Data via Diffusion Models (ICLR 2025 · similarity 0.86)
- Influence Functions for Scalable Data Attribution in Diffusion Models (ICLR 2025 · similarity 0.85)
- Revisiting Diffusion Model Predictions Through Dimensionality (NeurIPS 2026 · similarity 0.85)
- Distillation of Discrete Diffusion through Dimensional Correlations (ICML 2025 · similarity 0.85)
- Diffusion Models With Learned Adaptive Noise (NeurIPS 2024 · similarity 0.85)
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.
