Research map / Probabilistic and scientific modeling / Causal discovery
Amortized Inference for Causal Structure Learning
Lars Lorch, Scott Sussex, Jonas Rothfuss, Andreas Krause, Bernhard Schölkopf
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 AVICI instead of the strongest observational baseline | Causal structure learning · Linear, observational | SID: 215.6 ± 35.0 -> 145.6 ± 21.5 | Improved |
| Use AVICI instead of the strongest observational baseline | Causal structure learning · Rff, observational | F1: 0.476 ± 0.05 -> 0.618 ± 0.06 | Improved |
| Use AVICI instead of the strongest observational baseline | Causal structure learning · Grn, observational | SID: 588.7 ± 36.6 -> 641.7 ± 34.7 | Hurt |
| Use AVICI instead of the strongest interventional baseline | Causal structure learning · Linear, interventional | F1: 0.736 ± 0.03 -> 0.819 ± 0.02 | Improved |
| Use AVICI instead of the strongest interventional baseline | Causal structure learning · Rff, interventional | SID: 282.8 ± 46.3 -> 192.7 ± 44.8 | Improved |
| Use AVICI instead of the strongest interventional baseline | Causal structure learning · Grn, interventional | F1: 0.092 ± 0.01 -> 0.338 ± 0.06 | Improved |
| Increase network depth from 1 to 8 blocks (ablation) | AVICI · Rff in-distribution | SID: 267.2 ± 22.0 -> 65.2 ± 8.4 | Improved |
| Increase network depth from 2 to 8 blocks (ablation) | AVICI · Rff in-distribution | SID: 195.9 ± 18.5 -> 65.2 ± 8.4 | Improved |
| Increase network depth from 4 to 8 blocks (ablation) | AVICI · Rff in-distribution | SID: 116.6 ± 13.1 -> 65.2 ± 8.4 | Improved |
| Add attention over observation axis n (ablation) | AVICI · Rff in-distribution | SID: 351.5 ± 27.9 -> 65.2 ± 8.4 | Improved |
| Add attention over variable axis d (ablation) | AVICI · Rff in-distribution | SID: 416.8 ± 29.6 -> 65.2 ± 8.4 | Improved |
| Replace the inner-product theta model with a relational-network theta model (ablation) | AVICI · Rff | SID: 72.4 ± 9.2 vs. 65.2 ± 8.4; 225.7 ± 25.2 vs. 221.5 ± 24.7 | No clear effect |
Closest papers by meaning
- Learning Causal Structures Using Regression Invariance (NeurIPS 2017 · similarity 0.90)
- A Hybrid Causal Structure Learning Algorithm for Mixed-Type Data (AAAI 2022 · similarity 0.90)
- CausalTab: Pretraining Across Causal Environments for Tabular Causal Discovery (NeurIPS 2026 · similarity 0.89)
- Learning Structural Causal Models from Ordering: Identifiable Flow Models (AAAI 2025 · similarity 0.88)
- Learning Mixtures of Unknown Causal Interventions (NeurIPS 2024 · similarity 0.88)
- A Fixed-Point Approach for Causal Generative Modeling (ICML 2024 · similarity 0.88)
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.
