AI Paper Atlas

Research map / Probabilistic and scientific modeling / Treatment effects

Overlap-Adaptive Regularization for Conditional Average Treatment Effect Estimation

Valentyn Melnychuk, Dennis Frauen, Jonas Schweisthal, Stefan Feuerriegel

ICLR 2026 · Probabilistic and scientific modeling › Treatment effects · approach: Meta-learners and partial identification · 1 citations (OpenAlex)

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 changedWhereMeasuredEffect
add overlap-adaptive regularization (OAR) instead of constant regularizationDR-learner · HC-MNIST (noise 0.1)rPEHE_out: 0.741 -> 0.726Improved
add debiased OAR (dOAR) instead of constant regularizationDR-learner · HC-MNIST (dropout 0.1)rPEHE_out: 0.746 -> 0.713Improved
add dOAR instead of constant regularizationR-learner · HC-MNIST (dropout 0.1)rPEHE_out: 0.720 -> 0.689Improved
add OAR instead of constant regularizationR-learner · HC-MNIST (noise 0.25)rPEHE_out: 0.674 -> 0.673No clear effect
add OAR instead of constant regularizationR-learner · HC-MNIST (dropout 0.1)rPEHE_out: 0.720 -> 0.720No clear effect
add OAR instead of constant regularizationIVW-learner · HC-MNIST (dropout 0.5)rPEHE_out: 1.113 -> 0.997Improved
add dOAR instead of constant regularizationIVW-learner · HC-MNIST (dropout 0.5)rPEHE_out: 1.113 -> 1.027Improved
add dOAR instead of constant regularizationIVW-learner · HC-MNIST (dropout 0.3)rPEHE_out: 1.117 -> 1.110No clear effect
add debiasing (dOAR) to multiplicative OAR with dropout 0.3 (ablation)DR-learner · ACIC 2016% datasets significantly better than CR: 41.56% -> 70.13%Improved
add debiasing (dOAR) to multiplicative OAR with noise reg. 0.05 (ablation)DR-learner · ACIC 2016% datasets significantly better than CR: 31.17% -> 57.14%Improved
add debiasing (dOAR) to m^2 OAR with noise reg. 0.05 (ablation)DR-learner · ACIC 2016% datasets significantly better than CR: 27.27% -> 76.62%Improved
add debiasing (dOAR) to m^2 OAR with dropout 0.3 (ablation)DR-learner · ACIC 2016% datasets significantly better than CR: 16.88% -> 64.94%Improved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy