Research map / Probabilistic and scientific modeling / Causal discovery
Zero-shot causal learning
Hamed Nilforoshan, Michael Moor, Yusuf H Roohani, Yining Chen, Anja Šurina, Michihiro Yasunaga, Sara Oblak, Jure Leskovec
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 |
|---|---|---|---|
| remove meta-learning (train with ERM instead of Reptile) (ablation) | CaML · Claims, pancytopenia, unseen drugs | RATE @ 0.999: 0.39 -> 0.48 | Improved |
| remove meta-learning (train with ERM instead of Reptile) (ablation) | CaML · Claims, pancytopenia, unseen drugs | RATE @ 0.99: 0.11 -> 0.13 | No clear effect |
| replace RA-learner with an alternative pseudo-outcome estimator (ablation) | CaML · Claims, pancytopenia, unseen drugs | RATE @ 0.999: 0.45 -> 0.48 | Improved |
| replace RA-learner with an alternative pseudo-outcome estimator (ablation) | CaML · Claims, pancytopenia, unseen drugs | RATE @ 0.99: 0.14 -> 0.13 | No clear effect |
| remove meta-learning (train with ERM instead of Reptile) (ablation) | CaML · LINCS | PEHE 50 DEGs: 3.57 -> 3.56 | No clear effect |
| remove meta-learning (train with ERM instead of Reptile) (ablation) | CaML · LINCS | PEHE 20 DEGs: 3.79 -> 3.78 | No clear effect |
| replace RA-learner with an alternative pseudo-outcome estimator (ablation) | CaML · LINCS | PEHE 50 DEGs: 4.28 -> 3.56 | Improved |
| replace RA-learner with an alternative pseudo-outcome estimator (ablation) | CaML · LINCS | PEHE 20 DEGs: 4.60 -> 3.78 | Improved |
| use pseudo-outcomes (CaML) instead of meta-learning Y(w) and Y(0) directly (T-learner w/ meta-learning) | meta-learned CATE model · Claims, pancytopenia, unseen drugs | RATE @ 0.999: 0.40 -> 0.48 | Improved |
| use pseudo-outcomes (CaML) instead of meta-learning Y(w) and Y(0) directly (S-learner w/ meta-learning) | meta-learned CATE model · Claims, pancytopenia, unseen drugs | RATE @ 0.999: 0.21 -> 0.48 | Improved |
| zero-shot CaML versus RA-learner trained directly on test interventions | CaML · Claims, pancytopenia, unseen drugs | RATE @ 0.99: 0.14 vs 0.13 | No clear effect |
| zero-shot CaML versus X-learner trained on test interventions | CaML · Claims, pancytopenia, unseen drugs | RATE @ 0.999: 0.06 -> 0.48 | Improved |
Closest papers by meaning
- End-To-End Causal Effect Estimation from Unstructured Natural Language Data (NeurIPS 2024 · similarity 0.86)
- Towards Causal Foundation Model: on Duality between Optimal Balancing and Attention (ICML 2024 · similarity 0.86)
- Proximal Causal Inference With Text Data (NeurIPS 2024 · similarity 0.86)
- When Selection Meets Intervention: Additional Complexities in Causal Discovery (ICLR 2025 · similarity 0.85)
- Prediction-Powered Causal Inferences (NeurIPS 2025 · similarity 0.85)
- CausalPFN: Amortized Causal Effect Estimation via In-Context Learning (NeurIPS 2025 · similarity 0.84)
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.
