Research map / Reinforcement learning / Policy gradient and actor-critic
Uncertainty-Aware Reinforcement Learning for Risk-Sensitive Player Evaluation in Sports Game
Guiliang Liu, Yudong Luo, Oliver Schulte, Pascal Poupart
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 risk-aware distributional estimates (RiGIM at c=0.5) over risk-neutral GIM | GIM · NHL Assist correlation | correlation: 0.456 -> 0.675 | Improved |
| Add risk-aware distributional estimates (RiGIM at c=0.5) over risk-neutral GIM | GIM · Soccer Assist correlation | correlation: 0.462 -> 0.577 | Improved |
| Remove the epistemic uncertainty estimator (Na-RiGIM) (ablation) | RiGIM · NHL Point correlation | correlation: 0.678 -> 0.625 | Hurt |
| Remove the epistemic uncertainty estimator (Na-RiGIM) (ablation) | RiGIM · Soccer Key Passes correlation | correlation: 0.574 -> 0.503 | Hurt |
| Replace SP-CNF with GDA density estimator (ablation) | RiGIM · NHL TOI correlation | correlation: 0.68 -> 0.593 | Hurt |
| Replace SP-CNF with GDA density estimator (ablation) | RiGIM · Soccer Drb/KeyP correlation | correlation: 0.574 -> 0.478 | Hurt |
| Filter OoD events by low density in RiGIM (ablation) | RiGIM · NHL SHG correlation | correlation: 0.134 -> 0.11 | Hurt |
| Tune confidence level c on validation set (c*=0.34) instead of 0.5 | RiGIM · NHL Assist correlation | correlation: 0.675 -> 0.68 | No clear effect |
| Tune confidence level c on validation set (c*=0.49) instead of 0.5 | RiGIM · Soccer Goal correlation | correlation: 0.671 -> 0.682 | No clear effect |
| Make RiGIM more risk-seeking by lowering c | RiGIM · NHL success measures | correlation | Mixed |
| Replace tabular Scoring Impact with neural recurrent value models | SI · NHL Goal correlation | correlation: 0.596 -> 0.477 | Mixed |
| Use distribution estimates (RiGIM) versus expected values with FS-CNF filtering | RiGIM · Scoring chance match, bins | absolute difference | Improved |
Closest papers by meaning
- Pessimism Meets Risk: Risk-Sensitive Offline Reinforcement Learning (ICML 2024 · similarity 0.85)
- The impact of uncertainty on regularized learning in games (ICML 2025 · similarity 0.84)
- Dynamic Uncertainty Estimation for Offline Reinforcement Learning (AAAI 2025 · similarity 0.84)
- The CoachAI Badminton Environment: Bridging the Gap between a Reinforcement Learning Environment and Real-World Badminton Games (AAAI 2024 · similarity 0.84)
- The CoachAI Badminton Environment: A Novel Reinforcement Learning Environment with Realistic Opponents (Student Abstract) (AAAI 2024 · similarity 0.83)
- LATTE-MV: Learning to Anticipate Table Tennis Hits from Monocular Videos (CVPR 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.
