Research map / Reinforcement learning / Imitation learning
Variational Inverse Control with Events: A General Framework for Data-Driven Reward Definition
Justin Fu, Avi Singh, Dibya Ghosh, Larry Yang, Sergey Levine
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 |
|---|---|---|---|
| Use the ANY query instead of ALL | Event-based control · HalfCheetah | Min. Dist: 2.01 (0.48) -> 0.97 (0.46) | Improved |
| Use the ALL query instead of ANY | Event-based control · HalfCheetah | Avg. Dist: 1.35 (0.20) -> 1.33 (0.16) | No clear effect |
| Use the ANY query instead of ALL | Event-based control · Lobber | Min. Dist: 0.36 (0.21) -> 0.25 (0.20) | Improved |
| Use VICE-ALL event learning instead of the offline classifier | VICE · Maze | Final distance to goal: 0.35 (0.29) -> 0.20 (0.19) | Improved |
| Use VICE-ANY event learning instead of the offline classifier | VICE · Maze | Final distance to goal: 0.37 (0.21) -> 0.23 (0.15) | Improved |
| Use VICE-ALL event learning instead of the offline classifier | VICE · Ant | Final distance to goal: 2.71 (0.75) -> 0.64 (0.32) | Improved |
| Use VICE-ANY event learning instead of the offline classifier | VICE · Ant | Final distance to goal: 3.93 (1.56) -> 0.62 (0.55) | Improved |
| Use VICE-ALL event learning instead of the offline classifier | VICE · Pusher | Final distance to goal: 0.25 (0.01) -> 0.09 (0.01) | Improved |
| Use VICE-ANY event learning instead of the offline classifier | VICE · Pusher | Final distance to goal: 0.25 (0.01) -> 0.11 (0.01) | Improved |
| Use VICE-ALL instead of true binary event indicators | VICE · Ant | Final distance to goal: 1.61 (1.35) -> 0.64 (0.32) | Improved |
| Use VICE-ANY instead of true binary event indicators | VICE · Ant | Final distance to goal: 1.61 (1.35) -> 0.62 (0.55) | Improved |
| Use VICE-ALL instead of true binary event indicators | VICE · Pusher | Final distance to goal: 0.17 (0.03) -> 0.09 (0.01) | Improved |
Closest papers by meaning
- Is Inverse Reinforcement Learning Harder than Standard Reinforcement Learning? A Theoretical Perspective (ICML 2024 · similarity 0.89)
- Environment Design for Inverse Reinforcement Learning (ICML 2024 · similarity 0.89)
- Hybrid Inverse Reinforcement Learning (ICML 2024 · similarity 0.89)
- Recent Advancements in Inverse Reinforcement Learning (AAAI 2024 · similarity 0.89)
- A Unified Linear Programming Framework for Offline Reward Learning from Human Demonstrations and Feedback (ICML 2024 · similarity 0.88)
- How does Inverse RL Scale to Large State Spaces? A Provably Efficient Approach (NeurIPS 2024 · similarity 0.88)
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.
