Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Bridging the Imitation Gap by Adaptive Insubordination
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 adaptive weighting of imitation and reward-based RL losses | PoisonedDoors · PoisonedDoors | expected reward: -0.09 -> 1 | Helped |
| Use adaptive weighting of imitation and reward-based RL losses | ADVISOR · LavaCrossing Faulty Switch | expected reward: 0.04 -> 0.77 | Helped |
| Use adaptive weighting of imitation and reward-based RL losses | ADVISOR · LavaCrossing Once Switch | expected reward: 0 -> 0.8 | Helped |
| Use adaptive weighting of imitation and reward-based RL losses | ADVISOR · LavaCrossing Corrupt Exp. | expected reward: 0.74 -> 0.94 | Helped |
| Use adaptive weighting of imitation and reward-based RL losses | ADVISOR · PointGoal Navigation @10% | SPL: 48.9 -> 57.7 | Helped |
| Use adaptive weighting of imitation and reward-based RL losses | ADVISOR · PointGoal Navigation @100% | Success: 79.0 -> 88.2 | Helped |
| Use adaptive weighting of imitation and reward-based RL losses | ADVISOR · ObjectGoal Navigation @100% | Success: 31.6 -> 29.9 | Hurt |
| Use adaptive weighting of imitation and reward-based RL losses | ADVISOR · ObjectGoal Navigation @100% | SPL: 13.1 -> 14.1 | Helped |
| Use adaptive weighting of imitation and reward-based RL losses | ADVISOR · Cooperative Navigation @100% | Reward: $-$416.7 -> \textbf{$-$405.6} | Helped |
| Train with an expert whose observation privilege is closer to the student's | Behavior cloning on 2D-LH · 2D-LH | average episode length | Helped |
Closest papers by meaning
- Imitation Learning from Observation with Automatic Discount Scheduling (ICLR 2024 · similarity 0.90)
- Is Behavior Cloning All You Need? Understanding Horizon in Imitation Learning (NeurIPS 2024 · similarity 0.90)
- Interactive and Hybrid Imitation Learning: Provably Beating Behavior Cloning (NeurIPS 2025 · similarity 0.90)
- GO-DICE: Goal-Conditioned Option-Aware Offline Imitation Learning via Stationary Distribution Correction Estimation (AAAI 2024 · similarity 0.89)
- Learning When to Stop: Selective Imitation Learning Under Arbitrary Dynamics Shift (NeurIPS 2026 · similarity 0.89)
- EvIL: Evolution Strategies for Generalisable Imitation Learning (ICML 2024 · 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.