Research map / Reinforcement learning / Imitation learning
Interactive Learning from Activity Description
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 |
|---|---|---|---|
| train with activity descriptions instead of scalar rewards | ADEL versus the best REINFORCE baseline · Vision-language navigation | test success rate (%): +9.8% | Helped |
| train with activity descriptions instead of scalar rewards | ADEL versus REINFORCE · Word modification | test success rate (%): +88.1% | Helped |
| use activity descriptions rather than demonstrations | ADEL versus DAgger · Vision-language navigation | test success rate (%): 32.0 ± 1.63 -> 31.9 ± 0.76 | Hurt |
| use activity descriptions rather than demonstrations | ADEL versus DAgger · Word modification | test success rate (%): 93.0 ± 0.37 -> 89.0 ± 1.30 | Hurt |
| remove approximate marginal from execution sampling (ablation) | ADEL · Vision-language navigation | validation success rate (%): ~0.0 -> 32.0 | Helped |
| remove approximate marginal from execution sampling (ablation) | ADEL · Word modification | validation success rate (%): ~0.2 -> 88.0 | Helped |
| use only the approximate marginal for execution sampling (ablation) | ADEL · Vision-language navigation | validation success rate (%): 29.4 -> 32.0 | Hurt |
| use only the approximate marginal for execution sampling (ablation) | ADEL · Word modification | validation success rate (%): 55.7 -> 88.0 | Hurt |
| mix the approximate marginal with the agent-estimated conditional (ablation) | ADEL · Vision-language navigation | validation success rate (%): 29.4 -> 32.0 | Helped |
| mix the approximate marginal with the agent-estimated conditional (ablation) | ADEL · Word modification | validation success rate (%): 55.7 -> 88.0 | Helped |
Closest papers by meaning
- Agnostic Interactive Imitation Learning: New Theory and Practical Algorithms (ICML 2024 · similarity 0.88)
- When Does Interaction Help? Representational Tradeoffs in Value-Based Imitation Learning (NeurIPS 2026 · similarity 0.87)
- Interactive and Hybrid Imitation Learning: Provably Beating Behavior Cloning (NeurIPS 2025 · similarity 0.87)
- Imitation Learning from Observation with Automatic Discount Scheduling (ICLR 2024 · similarity 0.86)
- Provable Interactive Learning with Hindsight Instruction Feedback (ICML 2024 · similarity 0.86)
- Robot-Gated Interactive Imitation Learning with Adaptive Intervention Mechanism (ICML 2025 · similarity 0.86)
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.