Research map / Reinforcement learning / Imitation learning
Watch, Try, Learn: Meta-Learning from Demonstrations and Rewards
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 |
|---|---|---|---|
| Add a trial episode and condition the policy on both the demonstration and trial experience | WTL versus MIL · Gripper meta-test tasks (state-space) | success rate: $.30 \pm .02$ -> $.42 \pm .02$ | Helped |
| Use demonstration- and trial-conditioned meta-learning instead of behavior cloning without task conditioning | WTL versus BC · Gripper meta-test tasks (state-space) | success rate: $.09 \pm .01$ -> $.42 \pm .02$ | Helped |
| Use one demonstration and one trial with WTL instead of per-task SAC fine-tuning after behavior cloning | WTL versus BC + SAC · Gripper meta-test tasks (state-space) | success rate: WTL: $.42 \pm .02$; BC + SAC: $.11 \pm .07$, $.29 \pm .10$, $.39 \pm .11$ | Mixed |
Closest papers by meaning
- Meta-Imitation Learning by Watching Video Demonstrations (ICLR 2022 · similarity 0.92)
- Imitation Learning from Observation with Automatic Discount Scheduling (ICLR 2024 · similarity 0.91)
- RLIF: Interactive Imitation Learning as Reinforcement Learning (ICLR 2024 · similarity 0.89)
- Enhancing Online Reinforcement Learning with Meta-Learned Objective from Offline Data (AAAI 2025 · similarity 0.89)
- Uncertainty-Based Smooth Policy Regularisation for Reinforcement Learning with Few Demonstrations (NeurIPS 2025 · similarity 0.89)
- Learning from Demonstrations via Capability-Aware Goal Sampling (NeurIPS 2025 · 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.