Research map / Reinforcement learning / Imitation learning
Regularized Behavior Cloning for Blocking the Leakage of Past Action Information
Seokin Seo, HyeongJoo Hwang, Hongseok Yang, Kee-Eung Kim
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 |
|---|---|---|---|
| add HSCIC regularization on representation (PALR) | BC with observation history · hopper w=2 | normalized score: 32.5 -> 42.0 | Improved |
| add HSCIC regularization (PALR) | BC · walker2d w=2 | normalized score: 53.0 -> 79.8 | Improved |
| add HSCIC regularization (PALR) | BC · halfcheetah w=2 | normalized score: 74.1 -> 86.4 | Improved |
| add HSCIC regularization (PALR) | BC · ant w=4 | normalized score: 64.4 -> 64.6 | No clear effect |
| add HSCIC regularization (PALR) | BC · carla-lane | normalized score: 52.5 -> 72.9 | Improved |
| FCA adversarial conditional-entropy regularizer instead of none | BC · hopper w=4 | normalized score: 47.7 -> 36.9 | Hurt |
| MINE-based regularizer instead of none | BC · hopper w=2 | normalized score: 32.5 -> 25.0 | Hurt |
| RAP dual-stream method instead of BC | BC · walker2d w=2 | normalized score: 53.0 -> 15.8 | Hurt |
| regularize imitator action directly (PALR-ACT) instead of representation (ablation) | PALR · walker2d w=2 | normalized score: 79.83 -> 58.50 | Hurt |
| regularize imitator action directly (PALR-ACT) instead of representation (ablation) | PALR · ant w=4 | normalized score: 64.64 -> 65.00 | No clear effect |
| add VIB regularization (beta=10^-3) (ablation) | BC · walker2d w=2 | normalized score: 53.04 -> 44.28 | Hurt |
| PALR on Decision Transformer last hidden state (DT-PALR) | DT · walker2d | normalized score: 23.93 -> 32.58 | Improved |
Closest papers by meaning
- Is Behavior Cloning All You Need? Understanding Horizon in Imitation Learning (NeurIPS 2024 · similarity 0.90)
- Videos are Sample-Efficient Supervisions: Behavior Cloning from Videos via Latent Representations (NeurIPS 2025 · similarity 0.90)
- Learning When to Stop: Selective Imitation Learning Under Arbitrary Dynamics Shift (NeurIPS 2026 · similarity 0.89)
- Mitigating Covariate Shift in Behavioral Cloning via Robust Stationary Distribution Correction (NeurIPS 2024 · similarity 0.89)
- Diffusion Model-Augmented Behavioral Cloning (ICML 2024 · similarity 0.88)
- Imitation Learning from Observation with Automatic Discount Scheduling (ICLR 2024 · similarity 0.88)
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.
