Research map / Reinforcement learning / Imitation learning
Meta-Inverse Reinforcement Learning with Probabilistic Context Variables
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 PEMIRL instead of Meta-IL for one-shot task adaptation | one-shot policy generalization · Point Maze | return: -6.68 ± 1.51 vs -7.37 ± 1.02 | Mixed |
| use PEMIRL instead of Meta-InfoGAIL for one-shot task adaptation | one-shot policy generalization · Ant | return: 871.93 ± 31.28 vs 846.18 ± 25.81 | Hurt |
| use PEMIRL instead of Meta-InfoGAIL for one-shot task adaptation | one-shot policy generalization · Sweeper | return: -87.06 ± 6.57 vs -74.17 ± 5.31 | Helped |
| use PEMIRL instead of Meta-InfoGAIL for one-shot task adaptation | one-shot policy generalization · Sawyer Pusher | return: -27.56 ± 4.86 vs -27.16 ± 3.11 | Helped |
| infer a context-conditioned reward with PEMIRL rather than AIRL | reward adaptation · Point-Maze-Shift | return: -29.07 ± 4.12 vs -9.04 ± 1.09 | Helped |
| infer a context-conditioned reward with PEMIRL rather than Meta-InfoGAIL | reward adaptation · Disabled-Ant | return: -38.73 ± 6.41 vs 152.62 ± 11.75 | Helped |
| vary the latent dimension from 1 to 3 (ablation) | PEMIRL · Point-Maze-Shift | return: -10.58 ± 1.27 -> -14.13 ± 1.21 | Hurt |
| vary the latent dimension from 3 to 5 (ablation) | PEMIRL · Point-Maze-Shift | return: -14.13 ± 1.21 -> -15.41 ± 1.40 | Hurt |
| add the mutual information objective (ablation) | PEMIRL reward adaptation · Point-Maze-Shift | return: -39.24 ± 3.48 -> -14.13 ± 1.21 | Helped |
| use PEMIRL rather than Meta-IL in a stochastic environment | reward adaptation · Point-Maze-Shift stochastic | return: -30.58 ± 4.17 vs -17.39 ± 0.84 | Helped |
Closest papers by meaning
- Meta-Inverse Reinforcement Learning for Mean Field Games via Probabilistic Context Variables (AAAI 2024 · similarity 0.92)
- Rethinking Inverse Reinforcement Learning: from Data Alignment to Task Alignment (NeurIPS 2024 · similarity 0.91)
- Is Inverse Reinforcement Learning Harder than Standard Reinforcement Learning? A Theoretical Perspective (ICML 2024 · similarity 0.91)
- Recent Advancements in Inverse Reinforcement Learning (AAAI 2024 · similarity 0.90)
- Inverse Reinforcement Learning by Estimating Expertise of Demonstrators (AAAI 2025 · similarity 0.90)
- Offline Inverse RL: New Solution Concepts and Provably Efficient Algorithms (ICML 2024 · similarity 0.90)
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.