Research map / Reinforcement learning / Policy gradient and actor-critic
Target-Driven Policy Optimization for Sequential Counterfactual Outcome Control
Xin Wang, Xiangyu Zhang, Shengfei Lyu, Huanhuan Chen
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 |
|---|---|---|---|
| add reward rescaling (ablation) | GIFT · MIMIC-III (tau=3) | RMSE: 0.66 -> 0.33 | Improved |
| add reward rescaling (ablation) | GIFT · MIMIC-III (tau=6) | RMSE: 0.83 -> 0.39 | Improved |
| add reward rescaling (ablation) | GIFT · Tumor kappa=4 (tau=6) | RMSE: 0.42 -> 0.40 | Improved |
| add reward rescaling (ablation) | GIFT · Tumor kappa=2 (tau=6) | RMSE: 0.62 -> 0.14 | Improved |
| add hindsight experience replay (ablation) | GIFT · MIMIC-III (tau=6) | RMSE: 0.95 -> 0.39 | Improved |
| add hindsight experience replay (ablation) | GIFT · Tumor kappa=4 (tau=6) | RMSE: 1.53 -> 0.40 | Improved |
| add hindsight experience replay (ablation) | GIFT · Tumor kappa=2 (tau=3) | RMSE: 0.23 -> 0.11 | Improved |
| use reward rescaling instead of CQL (ablation) | GIFT · MIMIC-III (tau=3) | RMSE: 0.45 -> 0.33 | Improved |
| use reward rescaling instead of CQL (ablation) | GIFT · Tumor kappa=4 (tau=6) | RMSE: 1.04 -> 0.40 | Improved |
| add reward rescaling (variance reduction) (ablation) | GIFT · MIMIC-III (tau=3) std | RMSE: 0.14 to 0.39 | Improved |
| raise upper clip eps2 from tight (<=0.1) to range 1-10 (ablation) | GIFT · MIMIC and Tumor kappa=3 | RMSE: sweet spot in [1, 10], eps2 about 2 | Improved |
| loosen upper clip eps2 to >=20 (ablation) | GIFT · MIMIC and Tumor kappa=3 | RMSE: eps2 >= 20 | Hurt |
Closest papers by meaning
- Variational Counterfactual Intervention Planning to Achieve Target Outcomes (ICML 2025 · similarity 0.88)
- A Snapshot of Influence: A Local Data Attribution Framework for Online Reinforcement Learning (NeurIPS 2025 · similarity 0.86)
- Generative OOD-regularized Model-based Policy Optimization (NeurIPS 2026 · similarity 0.86)
- Foundation Policies with Hilbert Representations (ICML 2024 · similarity 0.86)
- Towards Safe Policy Learning under Partial Identifiability: A Causal Approach (AAAI 2024 · similarity 0.86)
- A General Framework for Off-Policy Learning with Partially-Observed Reward (ICLR 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.
