Research map / Probabilistic and scientific modeling / Causal discovery
Effect-Invariant Mechanisms for Policy Generalization
Sorawit Saengkyongam, Niklas Pfister, Predrag Klasnja, Susan Murphy, Jonas Peters
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 |
|---|---|---|---|
| test e-invariance instead of full invariance to find usable feature subsets | Wald invariance test · HeartSteps V1 | subsets not rejected at 5% level | Improved |
| use e-invariant policy instead of full-set policy for zero-shot generalization | R-learner policy · Augmented HeartSteps V1 (leave-one-user-out) | relative reward: 17 out of 27 users, p-value 0.008 | Improved |
| violate linear CATE assumption for the Wald test (ablation) | Wald e-invariance test · Simulated data (nonlinear main and treatment effect) | rejection rate of e-invariant set | Hurt |
| use Wald test instead of DR test when linear effects hold | e-invariance test · Simulated data (linear effects) | test power | Improved |
| make the main effect nonlinear for the Wald test (ablation) | Wald e-invariance test · Simulated data (nonlinear main effect) | test power | Hurt |
Closest papers by meaning
- Towards Safe Policy Learning under Partial Identifiability: A Causal Approach (AAAI 2024 · similarity 0.85)
- Horizon Generalization in Reinforcement Learning (ICLR 2025 · similarity 0.85)
- An Analysis of Causal Effect Estimation using Outcome Invariant Data Augmentation (NeurIPS 2025 · similarity 0.85)
- Learning Conditional Invariances through Non-Commutativity (ICLR 2024 · similarity 0.85)
- Towards Estimating Bounds on the Effect of Policies under Unobserved Confounding (NeurIPS 2024 · similarity 0.85)
- A Practical Guide for Incorporating Symmetry in Diffusion Policy (NeurIPS 2025 · similarity 0.85)
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.
