Research map / Probabilistic and scientific modeling / Causal discovery
Learning Quadratic Variance Function (QVF) DAG Models via OverDispersion Scoring (ODS)
Gunwoong Park, Garvesh Raskutti
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 l1-penalized likelihood regression instead of GES in ODS Step 1 (ablation) | Generalized ODS · Poisson and Binomial DAGs, p=10 and 100 | probability of recovering causal ordering | Improved |
| increase sample size n from 100 to 10000 (ablation) | Generalized ODS · Poisson and Binomial DAGs | probability of recovering causal ordering: n in {100, 500, 1000, 2500, 5000, 10000} | Improved |
| use ODS (l1 or GES-based) instead of MMHC and GES | DAG learning · Poisson and Binomial DAGs, p=10 and 100 | normalized Hamming distance (skeleton and directed edges) | Improved |
| add ODS ordering step on top of GES (ablation) | GES · Poisson and Binomial DAGs | Hamming distance to true DAG | Improved |
| small sample sizes for ODS | Generalized ODS · Poisson and Binomial DAGs | DAG recovery | Hurt |
| use ODS rather than MMHC in run-time | DAG learning · Poisson DAG, n=10000, p up to 100 | run-time | Hurt |
Closest papers by meaning
- CoLiDE: Concomitant Linear DAG Estimation (ICLR 2024 · similarity 0.88)
- Differentiable DAG Sampling (ICLR 2022 · similarity 0.87)
- Counting and Sampling from Markov Equivalent DAGs Using Clique Trees (AAAI 2019 · similarity 0.87)
- Learning Gaussian DAGs from Network Data (JMLR 2024 · similarity 0.87)
- Causal Discovery via Bayesian Optimization (ICLR 2025 · similarity 0.87)
- Effective Causal Discovery under Identifiable Heteroscedastic Noise Model (AAAI 2024 · 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.
