Research map / Reinforcement learning / Policy gradient and actor-critic
SHAP-Guided Kernel Actor-Critic for Explainable Reinforcement Learning
Na Li, Hangguan Shan, Wei Ni, Wenjie Zhang, Xinyu Li
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 learned RKHS-SHAP weighting to the actor-critic (vs Advanced AC without SHAP) (ablation) | RSA2C-KME · Pendulum-v1 (noise variance 0) | mean return: -719.00 -> -139.83 | Improved |
| use learned SHAP weights instead of uniform SHAP weights (ablation) | RSA2C-KME · Pendulum-v1 (noise variance 0) | mean return: -693.08 -> -139.83 | Improved |
| use the advanced SHAP-guided actor-critic instead of classical RKHS-AC (ablation) | RSA2C-KME · Pendulum-v1 (noise variance 0) | mean return: -899.85 -> -139.83 | Improved |
| add RKHS-SHAP weighting (vs Advanced AC) (ablation) | RSA2C-KME · BipedalWalker-v3 (noise variance 0) | mean return: 263.91 -> 279.18 | Improved |
| add RKHS-SHAP weighting (vs Advanced AC) (ablation) | RSA2C-KME · Ant-v5 (noise variance 0) | mean return: 953.62 -> 962.03 | Improved |
| use CME instead of KME for SHAP expectations | RSA2C · Pendulum-v1 (noise variance 0) | mean return: -139.83 -> -141.86 | No clear effect |
| use CME instead of KME for SHAP expectations | RSA2C · BipedalWalker-v3 (noise variance 0.01) | mean return: 272.96 -> 269.38 | No clear effect |
| increase state noise variance from 0 to 0.01 | RSA2C-KME · Pendulum-v1 | mean return: -139.83 -> -153.41 | Hurt |
| increase state noise variance from 0 to 0.01 | PPO · Ant-v5 | mean return: 677.48 -> -36.31 | Hurt |
| increase state noise variance from 0 to 0.01 | SAC · BipedalWalker-v3 | return standard deviation: 2.44 -> 82.09 | Hurt |
| use CME instead of KME (compute cost) | RSA2C · BipedalWalker-v3 | FLOPs per iteration: 0.381 GFLOPs -> 1.427 GFLOPs | Hurt |
| replace SAC (GPU) with RSA2C-KME (CPU) | RSA2C-KME · Ant-v5 | time-to-solution (s): 10287.5 -> 508.2 | Improved |
Closest papers by meaning
- ReLU to the Rescue: Improve Your On-Policy Actor-Critic with Positive Advantages (ICML 2024 · similarity 0.90)
- Decision-Aware Actor-Critic with Function Approximation and Theoretical Guarantees (NeurIPS 2023 · similarity 0.90)
- Non-Asymptotic Analysis for Single-Loop (Natural) Actor-Critic with Compatible Function Approximation (ICML 2024 · similarity 0.88)
- Actor-Free Continuous Control via Structurally Maximizable Q-Functions (NeurIPS 2025 · similarity 0.88)
- Value Improved Actor Critic Algorithms (NeurIPS 2025 · similarity 0.88)
- A Theoretical Justification for Asymmetric Actor-Critic Algorithms (ICML 2025 · 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.
