Research map / Probabilistic and scientific modeling / Causal discovery
Model-based Causal Discovery for Zero-Inflated Count Data
Junsouk Choi, Yang Ni
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 |
|---|---|---|---|
| increase sample size from 250 to 2000 | ZiG-DAG HC0 · Linear ZiG-DAG synthetic (d=50) | MCC: 0.660 -> 0.930 | Improved |
| use nonlinear instead of linear ZiG-DAG on nonlinear data (ablation) | ZiG-DAG · Nonlinear ZiG-DAG synthetic (n=500, d=10) | MCC: 0.596 -> 0.684 | Improved |
| use nonlinear instead of linear ZiG-DAG on nonlinear data (ablation) | ZiG-DAG · Nonlinear ZiG-DAG synthetic (n=500, d=10) | FDR: 0.377 -> 0.179 | Improved |
| initialize greedy search with the MRS DAG instead of empty graph (HC) (ablation) | ZiG-DAG HC · Linear ZiG-DAG synthetic (n=250, d=50) | MCC: 0.660 -> 0.713 | Improved |
| initialize greedy search with the MRS DAG instead of empty graph (HC) (ablation) | ZiG-DAG HC · Linear ZiG-DAG synthetic (n=1000, d=100) | MCC: 0.800 -> 0.795 | No clear effect |
| use tabu search instead of hill climbing from empty graph (ablation) | ZiG-DAG · Linear ZiG-DAG synthetic (n=1000, d=50) | MCC: 0.897 -> 0.903 | Improved |
| use a misspecified zero-inflated negative binomial instead of hyper-Poisson (ablation) | ZiG-DAG · Linear ZiG-DAG synthetic (n=1000, d=50) | TPR/FDR/MCC | No clear effect |
| apply ZiG-DAG to data without zero-inflation | ZiG-DAG-NB · Negative binomial BN (n=500, d=50) | MCC: 0.615 -> 0.712 | Improved |
| add a hidden confounder (sigma>0.5) to the data | ZiDAG · Latent confounder scenario (n=250) | ACC | Hurt |
| add a hidden confounder of increasing strength | ZiG-DAG · Latent confounder scenario (n=250) | ACC | No clear effect |
| compare ZiG-DAG with MRS and ZiDAG on real TF-target pairs | ZiG-DAG · scRNA-seq TF-target pairs (47 pairs) | accuracy: 51% -> 60% | Improved |
Closest papers by meaning
- Order-based structure learning for zero-inflated count data under data heterogeneity (NeurIPS 2026 · similarity 0.90)
- Bayesian Causal Structural Learning with Zero-Inflated Poisson Bayesian Networks (NeurIPS 2020 · similarity 0.90)
- Coarsening Linear Non-Gaussian Causal Models with Cycles (NeurIPS 2026 · similarity 0.87)
- Interventional Causal Discovery in a Mixture of DAGs (NeurIPS 2024 · similarity 0.87)
- Causal Discovery from Poisson Branching Structural Causal Model Using High-Order Cumulant with Path Analysis (AAAI 2024 · similarity 0.87)
- On the Identifiability of Poisson Branching Structural Causal Model Using Probability Generating Function (NeurIPS 2024 · similarity 0.86)
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.
