Research map / Reinforcement learning / Visual and representation RL
DOMINO: Decomposed Mutual Information Optimization for Generalized Context in Meta-Reinforcement Learning
Yao Mu, Yuzheng Zhuang, Fei Ni, Bin Wang, Jianyu Chen, Jianye HAO, Ping Luo
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 |
|---|---|---|---|
| replace entangled context (MINO) with decomposed MI optimization (DOMINO) (ablation) | PPO+DOMINO · Cartpole-f-l | test average return: 234 -> 283 | Improved |
| replace entangled context (MINO) with decomposed MI optimization (DOMINO) (ablation) | PPO+DOMINO · Pendulum-m-l | test average return: -526 -> -436 | Improved |
| replace entangled context (MINO) with decomposed MI optimization (DOMINO) (ablation) | PPO+DOMINO · Ant-m-d | test average return: 184 -> 216 | Improved |
| replace entangled context (MINO) with decomposed MI optimization (DOMINO) (ablation) | PPO+DOMINO · Halfcheetah-m-d | test average return: 824 -> 1034 | Improved |
| replace entangled context (MINO) with decomposed MI optimization (DOMINO) (ablation) | PPO+DOMINO · Slimhumanoid-m-d | test average return: 4275 -> 5258 | Improved |
| replace entangled context (MINO) with decomposed MI optimization (DOMINO) (ablation) | PPO+DOMINO · Hopper-m-d | test average return: 964 -> 1137 | Improved |
| replace entangled context (MINO) with decomposed MI optimization (DOMINO) (ablation) | PPO+DOMINO · Cripple-Ant-m-d | test average return: 113 -> 132 | Improved |
| replace entangled context (MINO) with decomposed MI optimization (DOMINO) (ablation) | PPO+DOMINO · Cripple-Halfcheetah-m-d | test average return: 1197 -> 1326 | Improved |
| use DOMINO context instead of PEARL context | PPO+DOMINO · Hopper-m-d | test average return: 874 -> 1137 | Improved |
| use DOMINO context instead of T-MCL context | PPO+DOMINO · Ant-m-d | test average return: 173 -> 216 | Improved |
| use DOMINO context instead of T-MCL context | PPO+DOMINO · Slimhumanoid-m-d | test average return: 4136 -> 5258 | Improved |
| PPO+DOMINO vs PEARL on training return | PPO+DOMINO · Cripple-Halfcheetah-m-d | test average return: 2538 -> 2503 | No clear effect |
Closest papers by meaning
- Towards an Information Theoretic Framework of Context-Based Offline Meta-Reinforcement Learning (NeurIPS 2024 · similarity 0.88)
- Decoupling Meta-Reinforcement Learning with Gaussian Task Contexts and Skills (AAAI 2024 · similarity 0.88)
- MetaCARD: Meta-Reinforcement Learning with Task Uncertainty Feedback via Decoupled Context-Aware Reward and Dynamics Components (AAAI 2024 · similarity 0.87)
- Meta-Reinforcement Learning with Universal Policy Adaptation: Provable Near-Optimality under All-task Optimum Comparator (NeurIPS 2024 · similarity 0.87)
- Skill-aware Mutual Information Optimisation for Zero-shot Generalisation in Reinforcement Learning (NeurIPS 2024 · similarity 0.87)
- Discovering Temporally-Aware Reinforcement Learning Algorithms (ICLR 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.
