Research map / Reinforcement learning / MDPs and planning
Jump Interval-Learning for Individualized Decision Making with Continuous Treatments
Hengrui Cai, Chengchun Shi, Rui Song, Wenbin Lu
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 |
|---|---|---|---|
| use deep neural network (D-JIL) instead of linear model (L-JIL) | JIL · Scenario 3 (p=20, n=800) | value function: 0.511 -> 0.545 | Improved |
| use linear model (L-JIL) instead of deep network (D-JIL) | JIL · Scenario 1 (p=20, n=800) | value function: 1.138 -> 1.297 | Improved |
| replace outcome-weighted learning (L-O-L) with L-JIL | L-O-L · Scenario 4 (p=20, n=800) | value function: 0.237 -> 0.882 | Improved |
| replace K-O-L with L-JIL | K-O-L · Scenario 2 (p=20, n=800) | value function: 0.276 -> 1.299 | Improved |
| replace Q-Linear with L-JIL | Q-Linear · Scenario 1 (p=20, n=800) | value function: 0.829 -> 1.297 | Improved |
| replace L-O-L with L-JIL | L-O-L · Scenario 5 (p=20, n=800) | value function: 7.81 -> 7.16 | Hurt |
| increase sample size from 50 to 800 | L-JIL · Scenario 1 (p=20) | value function: 0.783 -> 1.297 | Improved |
| increase sample size from 200 to 800 | L-JIL · Scenario 1 (p=4) | CI coverage probability (%): 89.80 -> 95.60 | Improved |
| increase sample size from 200 to 800 | L-JIL · Scenario 1 (p=4) | integrated l2 loss of varying coefficient: 0.371 -> 0.111 | Improved |
| use m=n/6 instead of m=n/10 (ablation) | L-JIL · Scenario 1 (p=20, n=800) | value function: 1.297 -> 1.289 | No clear effect |
| use m=n/6 instead of m=n/10 (ablation) | D-JIL · Scenario 1 (p=20, n=800) | value function: 1.138 -> 1.127 | No clear effect |
| use m=n/6 instead of m=n/10 (ablation) | D-JIL · Scenario 1 (p=20, n=800) | computation time (minutes): 35.48 -> 87.56 | Hurt |
Closest papers by meaning
- Stage-Aware Learning for Dynamic Treatments (JMLR 2024 · similarity 0.87)
- Efficient and Robust Transfer Learning of Optimal Individualized Treatment Regimes with Right-Censored Survival Data (JMLR 2025 · similarity 0.87)
- An effective framework for estimating individualized treatment rules (NeurIPS 2024 · similarity 0.87)
- Decision-Aware Proximal Bridge Learning for Optimal Treatment Selection (NeurIPS 2026 · similarity 0.84)
- Treatment Effect Estimation for Optimal Decision-Making (NeurIPS 2025 · similarity 0.84)
- Doubly Robust Fusion of Many Treatments for Policy Learning (ICML 2025 · similarity 0.83)
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.
