Research map / Reinforcement learning / Imitation learning
Deep Inverse Q-learning with Constraints
Gabriel Kalweit, Maria Huegle, Moritz Werling, Joschka Boedecker
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 inverse action-value iteration instead of MaxEnt IRL | Objectworld · Objectworld | EVD: 11.58 -> 0.09 | Improved |
| Use inverse Q-learning instead of MaxEnt IRL | Objectworld · Objectworld | EVD: 11.58 -> 1.47 | Improved |
| Use inverse action-value iteration instead of MaxEnt IRL | Objectworld · Objectworld | runtime: \SI{8.08}{\hour} -> \SI{1.77}{\minute} | Improved |
| Use inverse Q-learning instead of MaxEnt IRL | Objectworld · Objectworld | runtime: \SI{8.08}{\hour} -> \SI{21.06}{\minute} | Improved |
| Use a single inner value-iteration step in MaxEnt IRL (ablation) | MaxEnt IRL · Objectworld | runtime: \SI{8.08}{\hour} -> \SI{12.2}{\hour} | Hurt |
| Use a single inner value-iteration step in MaxEnt IRL (ablation) | MaxEnt IRL · Objectworld | EVD: 11.58 -> 4.33 | Improved |
| Add long-term keep-right and safety constraints to imitation | DCIQL · SUMO German highway scenario | constraint violations: 0 | Improved |
| Train on demonstrations with only 1000 samples | DCIQL · SUMO autonomous lane-changes | driving performance | Improved |
| Learn a reward through DCIQL rather than train CDQN on the true reward | SUMO autonomous lane-changes · SUMO highway scenarios | performance | Improved |
Closest papers by meaning
- Deep PQR: Solving Inverse Reinforcement Learning using Anchor Actions (ICML 2020 · similarity 0.90)
- Maximum Likelihood Constraint Inference for Inverse Reinforcement Learning (ICLR 2020 · similarity 0.90)
- Maximum-Likelihood Inverse Reinforcement Learning with Finite-Time Guarantees (NeurIPS 2022 · similarity 0.89)
- Learning Soft Constraints From Constrained Expert Demonstrations (ICLR 2023 · similarity 0.89)
- Is Inverse Reinforcement Learning Harder than Standard Reinforcement Learning? A Theoretical Perspective (ICML 2024 · 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.
