Research map / Reinforcement learning / Policy gradient and actor-critic
Off-Policy Evaluation via the Regularized Lagrangian
Mengjiao Yang, Ofir Nachum, Bo Dai, Lihong Li, Dale Schuurmans
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 dual estimator instead of primal estimator (ablation) | DICE regularized Lagrangian OPE · Grid, Reacher, Cartpole | OPE estimation error | Improved |
| use Lagrangian estimator instead of dual estimator (ablation) | DICE regularized Lagrangian OPE · Grid, Reacher, Cartpole | OPE estimation variance | Hurt |
| regularize dual variable instead of primal variable (with reward scaling on) (ablation) | dual estimator · Grid, Reacher, Cartpole | estimation bias | Improved |
| set reward scaling alpha_R=1 vs 0 under dual regularization (ablation) | dual estimator · Grid, Reacher, Cartpole | estimate accuracy | No clear effect |
| add normalization constraint (lambda) (ablation) | dual estimator with dual regularization · Grid, Reacher, Cartpole | training stability and approximation error | Improved |
| add positivity constraint on zeta (ablation) | dual estimator with dual regularization · Grid, Reacher, Cartpole | optimization stability and accuracy | Improved |
| solve unconstrained primal form instead of Lagrangian minimax (ablation) | dual estimator · Grid and Cartpole (discrete actions) | estimation variance/accuracy | Improved |
| solve unconstrained primal form instead of Lagrangian minimax (ablation) | dual estimator · Reacher (continuous actions) | estimation bias | Hurt |
| transform rewards by scaling or shifting | dual estimator vs primal/Lagrangian estimators · Grid, linear and neural parametrization | estimate robustness | Mixed |
Closest papers by meaning
- Relaxed Stationary Distribution Correction Estimation for Improved Offline Policy Optimization (AAAI 2024 · similarity 0.89)
- ODICE: Revealing the Mystery of Distribution Correction Estimation via Orthogonal-gradient Update (ICLR 2024 · similarity 0.88)
- Off-policy estimation with adaptively collected data: the power of online learning (NeurIPS 2024 · similarity 0.87)
- Efficient Policy Evaluation with Offline Data Informed Behavior Policy Design (ICML 2024 · similarity 0.87)
- Log-Sum-Exponential Estimator for Off-Policy Evaluation and Learning (ICML 2025 · similarity 0.87)
- Abstract Reward Processes: Leveraging State Abstraction for Consistent Off-Policy Evaluation (NeurIPS 2024 · similarity 0.87)
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.
