Research map / Reinforcement learning / Imitation learning
Offline Imitation Learning with Suboptimal Demonstrations via Relaxed Distribution Matching
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 asymmetrically relaxed support regularization instead of exact occupancy-distribution matching | Offline imitation learning with limited expert and suboptimal demonstrations · D4RL continuous-control and manipulation tasks, especially L3/L4 | normalized score: over 30% performance improvement on average over DemoDICE | Helped |
| Add density-ratio correction to RelaxDICE | RelaxDICE · D4RL tasks | normalized score: RelaxDICE-DRC achieves the best performance in 18 out of 22 tasks; RelaxDICE excels in the remaining 4 settings | Mixed |
| Vary regularization strength alpha (ablation) | RelaxDICE · halfcheetah and walker2d, four data settings each | normalized score | No clear effect |
| Vary regularization strength alpha (ablation) | DemoDICE · halfcheetah and walker2d, four data settings each | normalized score | Hurt |
| Use self-normalized importance weights for policy extraction rather than unnormalized weights (ablation) | RelaxDICE and RelaxDICE-DRC · D4RL tasks | normalized score | Helped |
Closest papers by meaning
- How to Leverage Diverse Demonstrations in Offline Imitation Learning (ICML 2024 · similarity 0.92)
- Imitation from Observations with Trajectory-Level Generative Embeddings (NeurIPS 2026 · similarity 0.92)
- Mitigating Covariate Shift in Behavioral Cloning via Robust Stationary Distribution Correction (NeurIPS 2024 · similarity 0.92)
- Is Behavior Cloning All You Need? Understanding Horizon in Imitation Learning (NeurIPS 2024 · similarity 0.91)
- Markov Balance Satisfaction Improves Performance in Strictly Batch Offline Imitation Learning (AAAI 2025 · similarity 0.90)
- SPRINQL: Sub-optimal Demonstrations driven Offline Imitation Learning (NeurIPS 2024 · similarity 0.90)
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.