Research map / Probabilistic and scientific modeling / Causal discovery
Identifiable Nonlinear Differentiable Causal Discovery via Independence and Adaptive Group Sparsity
Ruicong Yao, Tim Verdonck, Mihaela van der Schaar, Jakob Raymaekers
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 |
|---|---|---|---|
| minimize dependence of residuals and initialize from CAM (NOTIME-CAM) vs CAM alone | NOTIME-CAM · Sachs (real-world, n=853, d=11) | SHD: 30 -> 15 | Improved |
| use independence-based NOTIME-CAM instead of NOTEARS | NOTIME-CAM · Sachs (real-world, n=853, d=11) | SHD: 22 -> 15 | Improved |
| use NOTIME-CAM instead of DAGMA (default) | NOTIME-CAM · Sachs (real-world, n=853, d=11) | SHD: 17 -> 15 | Improved |
| tune hyperparameters to oracle values | NOTIME-CAM · Sachs (real-world, n=853, d=11) | SHD: 15 -> 14 | Improved |
| use MSE loss instead of independence measure | MSE-based direction test · Bivariate nonlinear ANM, Laplace noise | MSE: 1.101 vs 0.952 | Hurt |
| use NOTIME-CAM instead of NOTEARS and DAGMA on scaled data | NOTIME-CAM · Synthetic nonlinear ANM (d=10 to 70) | SHD | Improved |
| use NOTIME-CAM instead of CAM | NOTIME-CAM · Synthetic nonlinear ANM (d=10 to 70) | SHD | Improved |
| use NOTIME-CAM instead of NOTEARS on lognormal noise | NOTIME-CAM · Synthetic lognormal noise | SHD | No clear effect |
| use full proposal (CAM initialization plus adaptive group lasso) instead of ablation variants L1, glasso or CAM+L1 (ablation) | NOTIME-CAM · Synthetic ER graphs (Table 5) | SHD | Improved |
| vary thr_adapt (10% and 50% quantile times eta) (ablation) | NOTIME-CAM · Synthetic normal noise (Table 6) | SHD | No clear effect |
Closest papers by meaning
- Effective Causal Discovery under Identifiable Heteroscedastic Noise Model (AAAI 2024 · similarity 0.91)
- Strong and Weak Identifiability of Optimization-based Causal Discovery in Non-linear Additive Noise Models (ICML 2025 · similarity 0.90)
- Boosting Causal Structure Learning: An Asymmetric Exponential Modulation Gaussian-Based Adaptive Sample Reweighting Framework (AAAI 2025 · similarity 0.88)
- CoLiDE: Concomitant Linear DAG Estimation (ICLR 2024 · similarity 0.88)
- Differentiable Structure Learning with Ancestral Constraints (ICML 2025 · similarity 0.87)
- Differentiable Causal Discovery for Latent Hierarchical Causal Models (ICLR 2025 · 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.
