Research map / Reinforcement learning / Visual and representation RL
DataMIL: Selecting Data for Robot Imitation Learning with Datamodels
Shivin Dass, Alaa Khaddaj, Logan Engstrom, Aleksander Madry, Andrew Ilyas, Roberto Martín-Martín
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 DataMIL selection instead of BR retrieval | co-training with Octo · LIBERO-10 (avg) | success rate (%): 36.4 -> 37.76 | Improved |
| use DataMIL selection instead of Flow retrieval | co-training with Octo · Real-Average (OXE tasks) | success rate (%): 40.0 -> 61.0 | Improved |
| select data with DataMIL instead of using all data | Octo co-training · LIBERO-10 (avg) | success rate (%): 27.72 -> 37.76 | Improved |
| train All-Data baseline 2x longer (ablation) | All-Data · LIBERO-5 | success rate (%): 43.36 -> 44.13 | No clear effect |
| train All-Data baseline 4x longer (ablation) | All-Data · LIBERO-5 | success rate (%): 43.36 -> 38.40 | Hurt |
| use proxy loss metric instead of true rollouts (ablation) | DataMIL regression · MetaWorld (4 tasks) | success rate | No clear effect |
| add goal conditioning (ablation) | DataMIL · MetaWorld (50 tasks) | avg success | Improved |
| use finer clusters (sub-trajectory) instead of whole trajectories (ablation) | DataMIL · LIBERO-5 | success rate | Improved |
| select more prior data (>=20%) (ablation) | DataMIL vs BR/Flow · LIBERO-5 | success rate | No clear effect |
| select data with DataMIL instead of STRAP | co-training with Octo · Franka-Pouch (OXE-23) | success rate (%): 29.4 -> 70.6 | Improved |
Closest papers by meaning
- What Matters in Learning from Large-Scale Datasets for Robot Manipulation (ICLR 2025 · similarity 0.91)
- Data Scaling Laws in Imitation Learning for Robotic Manipulation (ICLR 2025 · similarity 0.90)
- DexFlyWheel: A Scalable and Self-improving Data Generation Framework for Dexterous Manipulation (NeurIPS 2025 · similarity 0.88)
- Robots Pre-train Robots: Manipulation-Centric Robotic Representation from Large-Scale Robot Datasets (ICLR 2025 · similarity 0.87)
- GraphMimic: Graph-to-Graphs Generative Modeling from Videos for Policy Learning (CVPR 2025 · similarity 0.86)
- Self-Improving Embodied Foundation Models (NeurIPS 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.
