Research map / Probabilistic and scientific modeling / Molecules and proteins
Learning Vine Copula Models for Synthetic Data Generation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 RL-based vine structure learning instead of the Gruber approach | vine fitting · Dense Vine | relative loglik(%): 81.0 -> 84.2 | Helped |
| use RL-based vine structure learning instead of Dißmann | vine fitting · Sparse Vine | relative loglik(%): 101.3 -> 100.2 | Hurt |
| use RL representation instead of vector representation (ablation) | vine model selection · Breast Cancer | log-likelihood per instance: 0.84 -> 0.91 | Helped |
| use RL representation instead of vector representation (ablation) | vine model selection · Wine | log-likelihood per instance: 0.037 -> 0.045 | Helped |
| use RL representation instead of vector representation (ablation) | vine model selection · Crime | log-likelihood per instance: 0.27 -> 0.31 | Helped |
| use RL Vine rather than Dißmann to generate synthetic training data | Decision Tree classifier · Breast Cancer | F1 score: 0.77 ± 0.12 -> 0.81 ± 0.06 | Helped |
| use RL Vine rather than Dißmann to generate synthetic training data | SVM classifier · Breast Cancer | F1 score: 0.55±0.05 -> 0.72 ± 0.03 | Helped |
| use RL Vine rather than Dißmann to generate synthetic training data | 5 layer MLP classifier · Breast Cancer | F1 score: 0.76 ± 0.06 -> 0.79 ± 0.03 | Helped |
| use RL Vine rather than Dißmann to generate synthetic training data | Decision Tree classifier · Wine Quality | F1 score: 0.13 ± 0.008 -> 0.21 ± 0.006 | Helped |
| use RL Vine rather than Dißmann to generate synthetic training data | SVM classifier · Wine Quality | F1 score: 0.09 ± 0.006 -> 0.09 ± 0.006 | No clear effect |
| use RL Vine rather than Dißmann to generate synthetic training data | 5 layer MLP classifier · Wine Quality | F1 score: 0.09 ± 0.006 -> 0.11 ± 0.008 | Helped |
| use RL Vine rather than Dißmann to generate synthetic training data | Decision Tree regressor · Crime | MSE: 0.137 ± 0.016 -> 0.096 ± 0.011 | Helped |
Closest papers by meaning
- Discrete Copula Diffusion (ICLR 2025 · similarity 0.83)
- Quasi-Bayes meets Vines (NeurIPS 2024 · similarity 0.81)
- Bridging Model-Based Optimization and Generative Modeling via Conservative Fine-Tuning of Diffusion Models (NeurIPS 2024 · similarity 0.81)
- Linear combinations of latents in generative models: subspaces and beyond (ICLR 2025 · similarity 0.80)
- Distribution Learning Meets Graph Structure Sampling (NeurIPS 2025 · similarity 0.80)
- A Fixed-Point Approach for Causal Generative Modeling (ICML 2024 · similarity 0.80)
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.