Research map / Reinforcement learning / MDPs and planning
State Abstraction as Compression in Apprenticeship Learning
David Abel, Dilip Arumugam, Kavosh Asadi, Yuu Jinnai, Michael L. Littman, Lawson L.S. Wong
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 |
|---|---|---|---|
| increase beta from 0 toward 1 (weight performance over compression) (ablation) | DIBS · Four Rooms | value relative to demonstrator | Improved |
| increase beta beyond 1 (ablation) | DIBS · Four Rooms | policy value | No clear effect |
| use stochastic abstraction (SIBS) instead of deterministic (DIBS) (ablation) | state abstraction · Four Rooms | policy value at low beta | Hurt |
| increase beta (ablation) | VAE-based SIBS · Breakout | reconstruction quality and reward | Improved |
Closest papers by meaning
- Apprenticeship Learning via Frank-Wolfe (AAAI 2020 · similarity 0.86)
- Deciding What to Learn: A Rate-Distortion Approach (ICML 2021 · similarity 0.86)
- State Abstractions for Lifelong Reinforcement Learning (ICML 2018 · similarity 0.85)
- Near Optimal Behavior via Approximate State Abstraction (ICML 2016 · similarity 0.85)
- When Do Skills Help Reinforcement Learning? A Theoretical Analysis of Temporal Abstractions (ICML 2024 · similarity 0.85)
- Building Minimal and Reusable Causal State Abstractions for Reinforcement Learning (AAAI 2024 · similarity 0.85)
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.
