Research map / Reinforcement learning / Imitation learning
Robust Imitation via Mirror Descent Inverse Reinforcement Learning
Dong-Sig Han, Hyunseo Kim, Hyundo Lee, JeHwan Ryu, Byoung-Tak Zhang
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 |
|---|---|---|---|
| replace RAIRL with MD-AIRL (Shannon) | RAIRL · bandit |A|=10^3 | Bregman divergence x |A|: 140.3 -> 125.3 | Improved |
| replace RAIRL with MD-AIRL (Tsallis) | RAIRL · bandit |A|=10^3 | Bregman divergence x |A|: 0.55 -> 0.24 | Improved |
| replace RAIRL with MD-AIRL (Tsallis) | RAIRL · bandit |A|=10^4 | Bregman divergence x |A|: 4.95 -> 4.21 | Improved |
| replace RAIRL with MD-AIRL (sin) | RAIRL · bandit |A|=10^3 | Bregman divergence x |A|: 3.82 -> 1.07 | Improved |
| replace RAIRL with MD-AIRL (Shannon) | RAIRL · bandit |A|=10^2 | Bregman divergence x |A|: 2.55 -> 2.28 | Improved |
| replace RAIRL with MD-AIRL (exp) | RAIRL · bandit |A|=10^4 | Bregman divergence x |A|: 5.06 -> 4.97 | No clear effect |
| replace RAIRL with MD-AIRL (Tsallis) | RAIRL · Walker2d noisy demos eps=0.01 | score: 2731.84 -> 3624.00 | Improved |
| replace RAIRL with MD-AIRL (Shannon) | RAIRL · Walker2d noisy demos eps=0.5 | score: 2451.00 -> 3252.65 | Improved |
| replace RAIRL with MD-AIRL (Shannon) | RAIRL · Ant noisy demos eps=0.5 | score: 3777.78 -> 4284.38 | Improved |
| replace RAIRL with MD-AIRL (Tsallis) | RAIRL · HalfCheetah noisy demos eps=0.5 | score: 4216.67 -> 4247.44 | Improved |
| replace RAIRL with MD-AIRL (Shannon) | RAIRL · Hopper noisy demos eps=0.01 | score: 3636.03 -> 3669.25 | No clear effect |
| replace RAIRL with MD-AIRL | RAIRL · MuJoCo, 32 configurations | average score: 30 of 32 | Improved |
Closest papers by meaning
- Mirror Descent Beyond Euclidean Stability: An Exponential Separation in Initialization Sensitivity (NeurIPS 2026 · similarity 0.88)
- Rethinking Inverse Reinforcement Learning: from Data Alignment to Task Alignment (NeurIPS 2024 · similarity 0.88)
- Non-Adversarial Inverse Reinforcement Learning via Successor Feature Matching (ICLR 2025 · similarity 0.88)
- A Geometric Perspective on Reward Function Updates in Inverse Reinforcement Learning (NeurIPS 2026 · similarity 0.87)
- Trust Region Reward Optimization and Proximal Inverse Reward Optimization Algorithm (NeurIPS 2025 · similarity 0.87)
- Recent Advancements in Inverse Reinforcement Learning (AAAI 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.
