AI Paper Atlas

Research map / Probabilistic and scientific modeling / Treatment effects

Deep Jump Learning for Off-Policy Evaluation in Continuous Treatment Settings

Hengrui Cai, Chengchun Shi, Rui Song, Wenbin Lu

NeurIPS 2021 · Probabilistic and scientific modeling › Treatment effects · approach: Benchmarking treatment effect models · 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
Use adaptive discretization of treatment space with DJL instead of a single-bandwidth kernelDR value estimator · Toy example V2 (pi(X)>0.5)absolute error: 1.09 -> 0.09Improved
Use adaptive discretization instead of a single large bandwidth h=1value estimator · Toy example V1absolute error: 0.40 -> 0.31Improved
Use small bandwidth h=0.4 versus h=1 in kernel estimatorkernel-based OPE · Toy example V1 and V2absolute error: 0.40 -> 0.50Mixed
Replace kernel baselines with DJLpolicy value estimator · Warfarin calibrated datamean squared error: 0.943 -> 0.240Improved
Replace SLOPE with DJLpolicy value estimator · Warfarin calibrated dataabsolute error: 0.611 -> 0.259Improved
Increase sample size from 50 to 300DJL · Scenario 1absolute error: 0.445 -> 0.209Improved
Increase sample size from 50 to 300DJL · Scenario 3absolute error: 2.014 -> 1.267Improved
Use kernel baseline 1 instead of DJLkernel OPE · Scenario 3, n=300absolute error: 1.267 -> 1.151Improved
Compare DJL with SLOPE at small sample sizeDJL · Scenario 1, n=50absolute error: 0.445 -> 0.392No clear effect
Replace DJL with SLOPEpolicy value estimator · Scenario 2, n=300absolute error: 0.411 -> 1.209Hurt
Choose number of initial intervals m=n/10 (ablation)DJL · Scenario 1absolute error and computational costImproved
Replace first kernel baseline with DJLDJL · Scenario 1, n=50 vs DJL n=300minutes: 365 -> 14Improved

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