Research map / Reinforcement learning / Visual and representation RL
Near Optimal Behavior via Approximate State Abstraction
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 approximate Q*-based state abstraction with increasing ε | NChain · NChain | abstract policy value | No clear effect |
| Use approximate Q*-based state abstraction with increasing ε | Minefield · Minefield | abstract policy value | No clear effect |
| Use approximate Q*-based state abstraction with ε between $.02 and $.025 | Taxi · Taxi | abstract policy value | No clear effect |
| Increase ε beyond $.025 | Taxi with approximate Q*-based abstraction · Taxi | abstract policy value | Hurt |
| Increase ε | Random MDP with approximate Q*-based abstraction · Random MDP | abstract policy value | Hurt |
| Use exact Q*-based abstraction (ε = 0) | Upworld · Upworld | optimal policy value | No clear effect |
| Use exact Q*-based abstraction (ε = 0) | Random MDP · Random MDP | abstract state count | No clear effect |
| Aggregate states whose optimal Q-values differ by at most ε | MDPs · MDPs | ground-MDP policy suboptimality: 2ε/(1-γ)^2 | Helped |
| Aggregate states with rewards and transitions within ε | MDPs · MDPs | ground-MDP policy suboptimality: (2ε + 2γε (|S_G| - 1))/(1-γ)^3 | Helped |
Closest papers by meaning
- State Abstractions for Lifelong Reinforcement Learning (ICML 2018 · similarity 0.90)
- Autonomous Option Invention for Continual Hierarchical Reinforcement Learning and Planning (AAAI 2025 · similarity 0.89)
- Oracle-Efficient Reinforcement Learning for Max Value Ensembles (NeurIPS 2024 · similarity 0.87)
- Reconciling Spatial and Temporal Abstractions for Goal Representation (ICLR 2024 · similarity 0.87)
- Reward-Respecting Subtasks for Model-Based Reinforcement Learning (Abstract Reprint) (AAAI 2024 · similarity 0.87)
- Polynomial-Time Approximability of Constrained Reinforcement Learning (ICML 2025 · similarity 0.87)
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.