Research map / Reinforcement learning / Imitation learning
TaSIL: Taylor Series Imitation Learning
Daniel Pfrommer, Thomas TCK Zhang, Stephen Tu, Nikolai Matni
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 higher-order p-TaSIL loss (p=1,2) instead of p=0 (standard imitation loss) | imitation learning on delta-ISS system · synthetic delta-ISS system | imitation gap and mean policy discrepancy | Improved |
| slower-decaying class K stability functions (small nu) | 1-TaSIL and 2-TaSIL · synthetic delta-ISS system | imitation gap | Improved |
| add 1-TaSIL loss to Behavior Cloning, DAgger and DART | BC, DAgger, DART · MuJoCo Walker2d, HalfCheetah, Humanoid, Ant | expert-normalized reward vs trajectory budget | Improved |
| add 1-TaSIL loss | BC, DART, DAgger · MuJoCo Swimmer | expert-normalized reward | No clear effect |
| off-policy 1-TaSIL Behavior Cloning versus on-policy DAgger and DART | BC with 1-TaSIL · MuJoCo tasks | expert-normalized reward | Improved |
| finite-difference approximation instead of Jacobian-based 1-TaSIL (ablation) | 1-TaSIL · MuJoCo tasks | expert-normalized reward | No clear effect |
Closest papers by meaning
- Interactive and Hybrid Imitation Learning: Provably Beating Behavior Cloning (NeurIPS 2025 · similarity 0.90)
- Split the Differences, Pool the Rest: Provably Efficient Multi-Objective Imitation (NeurIPS 2026 · similarity 0.90)
- Is Behavior Cloning All You Need? Understanding Horizon in Imitation Learning (NeurIPS 2024 · similarity 0.89)
- Learning When to Stop: Selective Imitation Learning Under Arbitrary Dynamics Shift (NeurIPS 2026 · similarity 0.89)
- Causal Imitation for Markov Decision Processes: a Partial Identification Approach (NeurIPS 2024 · similarity 0.89)
- Agnostic Interactive Imitation Learning: New Theory and Practical Algorithms (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.
