Research map / Reinforcement learning / Policy gradient and actor-critic
Model-based Adversarial Meta-Reinforcement Learning
Zichuan Lin, Garrett Thomas, Guangwen Yang, Tengyu Ma
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 |
|---|---|---|---|
| train on adversarially chosen tasks (AdMRL) instead of uniformly sampled tasks for meta-training (ablation) | model-based meta-RL (SLBO) · Ant2D 6x6 test grid, boundary tasks | sub-optimality gap | Improved |
| adversarial task selection instead of Gaussian task sampling (ablation) | model-based meta-RL (SLBO) · Ant2D worst-case tasks | worst-case sub-optimality gap | Improved |
| adversarial task selection instead of uniform/Gaussian sampling for high-dimensional task parameters (ablation) | model-based meta-RL (SLBO) · Cheetah-Highdim | worst-case sub-optimality gap | Improved |
| adversarial task selection instead of uniform/Gaussian sampling, tested on out-of-distribution tasks (ablation) | model-based meta-RL (SLBO) · Ant2D OOD tasks in [-5,5]^2 | sub-optimality gap | Improved |
| adversarial task selection instead of uniform/Gaussian sampling (ablation) | learned dynamics model · Ant2D OOD tasks | model prediction L2 error | Improved |
| use AdMRL instead of MAML and PEARL | meta-RL on MuJoCo velocity-control tasks · Hopper2D, Walker2D, Ant2D, Ant3D, Cheetah-Highdim | average return A_n | Improved |
| give MAML and PEARL a multi-task policy conditioned on task parameters | MAML, PEARL · MuJoCo velocity-control tasks | average return | No clear effect |
Closest papers by meaning
- Meta-Reinforcement Learning in Non-Stationary and Dynamic Environments (TPAMI 2022 · similarity 0.90)
- AMAGO-2: Breaking the Multi-Task Barrier in Meta-Reinforcement Learning with Transformers (NeurIPS 2024 · similarity 0.90)
- MAMBA: an Effective World Model Approach for Meta-Reinforcement Learning (ICLR 2024 · similarity 0.90)
- Meta Reinforcement Learning with Finite Training Tasks - a Density Estimation Approach (NeurIPS 2022 · similarity 0.90)
- Meta-Reinforcement Learning with Universal Policy Adaptation: Provable Near-Optimality under All-task Optimum Comparator (NeurIPS 2024 · similarity 0.90)
- MetaCARD: Meta-Reinforcement Learning with Task Uncertainty Feedback via Decoupled Context-Aware Reward and Dynamics Components (AAAI 2024 · similarity 0.89)
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.
