Research map / Probabilistic and scientific modeling / Causal discovery
Structure Learning of Undirected Graphical Models for Count Data
Nguyen Thi Kim Hue, Monica Chiogna
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 PC-LPGM conditional-independence testing instead of LPGM penalized regression | structure learner · Poisson random graph, p=100, pi=0.05, lambda_noise=0.5 | PPV: 0.337 -> 0.961 | Improved |
| use PC-LPGM instead of LPGM | structure learner · Poisson random graph, p=10, pi=0.1, lambda_noise=0.5 | runtime (seconds): 8.372 -> 1.577 | Improved |
| use PC-LPGM instead of LPGM | structure learner · Poisson random graph, p=100, pi=0.01, lambda_noise=0.5 | runtime (seconds): 81.192 -> 8.908 | Improved |
| use PC-LPGM instead of LPGM | structure learner · Poisson random graph, p=100, pi=0.02, lambda_noise=0.5 | sensitivity: 0.096 -> 0.946 | Improved |
| use PC-LPGM instead of LPGM on dense graphs | structure learner · Poisson random graph, p=10, pi=0.9, lambda_noise=0.5 | sensitivity: 0.827 -> 0.385 | Hurt |
| scale predictor X1 up by factor 5 (LPGM vs PC-LPGM) | LPGM · Variable scaling, p=10, pi=0.3 | sensitivity: 0.946 -> 0.082 | Hurt |
| scale predictor X1 up by factor 5 (PC-LPGM) | PC-LPGM · Variable scaling, p=10, pi=0.3 | sensitivity: 0.911 -> 0.992 | No clear effect |
| use truncated Poisson model (PC-TPGM) instead of Poisson (PC-LPGM) | PC-LPGM · random graphs, p=10,50,100, n=1000 | runtime | Hurt |
| use PC-TPGM instead of TPGM | truncated Poisson learner · random graphs, p=10,50,100, n=1000 | runtime | Improved |
| use PC-LPGM instead of Gaussian-based VSL and GLASSO | structure learner · Poisson graphs p=10,100, n=200-2000 | TP, PPV, Se | Improved |
| replace Gaussian assumption with nonparanormal (NPN-Copula, NPN-Skeptic) | Gaussian-based learners · Poisson graphs p=10,100 | structure recovery | No clear effect |
| increase conditioning-set size m (exponential cost) (ablation) | PC-LPGM · Poisson graph, p large | runtime | Hurt |
Closest papers by meaning
- Structure Learning with Side Information: Sample Complexity (NeurIPS 2019 · similarity 0.87)
- Learning to Discover Sparse Graphical Models (ICML 2017 · similarity 0.86)
- A Generic Family of Graphical Models: Diversity, Efficiency, and Heterogeneity (ICML 2025 · similarity 0.84)
- Gaussian Processes on Cellular Complexes (ICML 2024 · similarity 0.84)
- Distribution Learning Meets Graph Structure Sampling (NeurIPS 2025 · similarity 0.84)
- Graph Structure Inference with BAM: Neural Dependency Processing via Bilinear Attention (NeurIPS 2024 · similarity 0.83)
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.
