Research map / Reinforcement learning / Visual and representation RL
Interpretable Reward Redistribution in Reinforcement Learning: A Causal Approach
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 causal reward redistribution and compact state representation instead of the strongest baseline | GRD · Half-Cheetah | average accumulative reward: 1.25 \times 10^4 -> 1.5 \times 10^4 | Helped |
| use causal reward redistribution and compact state representation instead of the strongest baseline | GRD · Ant | average accumulative reward: 6 \times 10^3 -> 6.5 \times 10^3 | Helped |
| use the compact representation instead of all state dimensions as policy input (ablation) | GRD · MuJoCo tasks | average accumulative reward | Helped |
| increase λ1 from 5e-6 to 5e-5, making the learned reward-causal structure sparser (ablation) | GRD · Swimmer | average accumulative reward: 262± 94(0.66) -> 158± 90(0.17) | Hurt |
| increase λ1 from 0 to 3e-6 (ablation) | GRD · Swimmer | average accumulative reward: 159± 25(0.78) -> 220±102(0.50) | Helped |
| inject Gaussian noise into state dimensions excluded from the compact representation (ablation) | GRD · Ant | evaluation performance | No clear effect |
| replace SAC with DDPG or TD3 as the policy-learning backbone (ablation) | GRD · Half-Cheetah | average accumulative reward | Helped |
Closest papers by meaning
- Correlated Proxies: A New Definition and Improved Mitigation for Reward Hacking (ICLR 2025 · similarity 0.88)
- The Distributional Reward Critic Framework for Reinforcement Learning Under Perturbed Rewards (AAAI 2025 · similarity 0.88)
- Hierarchical Reinforcement Learning with Targeted Causal Interventions (ICML 2025 · similarity 0.88)
- Reward-Aware Proto-Representations in Reinforcement Learning (NeurIPS 2025 · similarity 0.88)
- On the Expressivity of Objective-Specification Formalisms in Reinforcement Learning (ICLR 2024 · similarity 0.87)
- The Value of Reward Lookahead in Reinforcement Learning (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.