Research map / Probabilistic and scientific modeling / Treatment effects
A Distillation Approach to Data Efficient Individual Treatment Effect Estimation
Maggie Makar, Adith Swaminathan, Emre Kıcıman
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 |
|---|---|---|---|
| distill GRF into a feature-budgeted decision tree (DEITEE) | GRF · ACIC semi-synthetic | features required: 58 -> less than 5 | Improved |
| distill BART into a decision tree (DEITEE) | BART · ACIC semi-synthetic | features required: 88% reduction | Improved |
| distill GRF with DEITEE | GRF · ACIC poly low exp high | MSE-T: 11.12 -> 11.63 | No clear effect |
| distill BART with DEITEE | BART · ACIC poly low exp high | MSE-T: 7.31 -> 8.97 | Hurt |
| distill GRF with DEITEE | GRF · ACIC step low step low | MSE-T: 3.68 -> 3.75 | No clear effect |
| distill BART with DEITEE | BART · ACIC step low step low | MSE-T: 1.56 -> 2.07 | Hurt |
| exponential response surface instead of step or linear | DEITEE · ACIC | MSE-T | Hurt |
| DEITEE distillation versus retraining with top-k important variables (RT models) | DEITEE-GRF and DEITEE-BART · ACIC hard setup | MSE and median SE | Improved |
| DEITEE versus retraining at the error plateau | DEITEE · ACIC hard setup | MSE | Hurt |
| DEITEE versus retraining at the error plateau | DEITEE · ACIC hard setup | median SE | No clear effect |
| distill BART for smoking effect | BART · CDC birth weight, smoking | MAE-P: 580.20 -> 580.20 | No clear effect |
| distill GRF for smoking effect | GRF · CDC birth weight, smoking | MAE-P: 581.27 -> 581.38 | No clear effect |
Closest papers by meaning
- Continuous Treatment Effect Estimation Using Gradient Interpolation and Kernel Smoothing (AAAI 2024 · similarity 0.87)
- PairNet: Training with Observed Pairs to Estimate Individual Treatment Effect (ICML 2024 · similarity 0.87)
- A Non-parametric Direct Learning Approach to Heterogeneous Treatment Effect Estimation under Unmeasured Confounding (NeurIPS 2024 · similarity 0.87)
- Estimating Heterogeneous Treatment Effects by Combining Weak Instruments and Observational Data (NeurIPS 2024 · similarity 0.87)
- An effective framework for estimating individualized treatment rules (NeurIPS 2024 · similarity 0.86)
- AutoCATE: End-to-End, Automated Treatment Effect Estimation (ICML 2025 · similarity 0.86)
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.
