Research map / Reinforcement learning / Offline RL
Distributional Offline Policy Evaluation with Predictive Error Guarantees
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 Gaussian mixture MLE (GMM-FLE) instead of quantile regression TD for return distribution estimation | FLE in 1-d combination lock · 1-d combination lock H=20 | approx. d_tv to true return distribution | Helped |
| Use diffusion model MLE (Diff-FLE) in place of quantile regression TD | FLE in 1-d combination lock · 1-d combination lock H=20 | approx. d_tv to true return distribution | Helped |
| Use diffusion model MLE (Diff-FLE) in place of 2-d categorical algorithm with 900 atoms | FLE in 2-d reward combination lock · 2-d reward combination lock H=10 | approx. d_tv to true return distribution: 0.357 vs 0.483 at h=1; 0.207 vs 0.453 at h=9 | Helped |
| Use Gaussian mixture MLE (GMM-FLE) in place of diffusion model in 2-d reward setting | FLE in 2-d reward combination lock · 2-d reward combination lock H=10 | approx. d_tv to true return distribution: 0.502 vs 0.207 at h=9 | Hurt |
| Use diffusion model MLE (Diff-FLE) instead of Gaussian mixture MLE in the 1-d reward setting | FLE in 1-d combination lock · 1-d combination lock H=20 | approx. d_tv to true return distribution: 0.292 vs 0.039 at h=1 | Hurt |
Closest papers by meaning
- A Principled Path to Fitted Distributional Evaluation (NeurIPS 2025 · similarity 0.87)
- Learning from Sparse Offline Datasets via Conservative Density Estimation (ICLR 2024 · similarity 0.86)
- OPERA: Automatic Offline Policy Evaluation with Re-weighted Aggregates of Multiple Estimators (NeurIPS 2024 · similarity 0.86)
- Generative OOD-regularized Model-based Policy Optimization (NeurIPS 2026 · similarity 0.86)
- Off-policy Evaluation Beyond Overlap: Sharp Partial Identification Under Smoothness (ICML 2024 · similarity 0.85)
- Efficient and Sharp Off-Policy Evaluation in Robust Markov Decision Processes (NeurIPS 2024 · 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.