Research map / Probabilistic and scientific modeling / Treatment effects
Exploratory Causal Inference in SAEnce
Tommaso Mencattini, Riccardo Cadei, Francesco Locatello
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 or effect magnitude under standard multiple testing (t-test, FDR, Bonferroni) | baseline testing procedures · CelebA semi-synthetic RCT (r=2) | Precision / IoU | Hurt |
| Use Neural Effect Search (recursive stratification) instead of vanilla multiple testing | SAE codes on SigLIP features · CelebA semi-synthetic RCT (r=2) | Precision / Recall / IoU | Improved |
| Use NES when the number of effects r is known | NES vs baselines · CelebA semi-synthetic RCT (r=2) | Precision and Recall: baselines stay below 0.5 on Precision and Recall | Improved |
| Use plain t-test or top-k instead of NES at very small sample (n=30) and effect (tau=0.1) | testing procedures · CelebA semi-synthetic RCT | Precision | No clear effect |
| Skip Bonferroni adjustment because of small sample size (ablation) | NES · ISTAnt real trial (n=44 videos) | number of discovered neurons: two neurons returned | Improved |
Closest papers by meaning
- Smoke and Mirrors in Causal Downstream Tasks (NeurIPS 2024 · similarity 0.89)
- CausalPFN: Amortized Causal Effect Estimation via In-Context Learning (NeurIPS 2025 · similarity 0.88)
- Towards Learning and Explaining Indirect Causal Effects in Neural Networks (AAAI 2024 · similarity 0.87)
- End-To-End Causal Effect Estimation from Unstructured Natural Language Data (NeurIPS 2024 · similarity 0.87)
- Causal Representation Learning via Counterfactual Intervention (AAAI 2024 · similarity 0.87)
- A Neural Framework for Generalized Causal Sensitivity Analysis (ICLR 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.
