Research map / Reinforcement learning / MDPs and planning
Kinematic State Abstraction and Provably Efficient Rich-Observation Reinforcement Learning
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 HOMER's learned kinematic abstraction and strategic policy-cover exploration | Diabolical combination lock · Diabolical combination lock, horizons tested | maximum horizon solved | Helped |
| Use naive entropy-bonus exploration instead of strategic exploration | A2C and PPO · Diabolical combination lock, H>3 | problem solved | Hurt |
| Add a Random Network Distillation exploration bonus | A2C and PPO · Diabolical combination lock, horizons up to 25 versus 50 and 100 | maximum horizon solved | Mixed |
| Use PCID's model-based policy cover with policies defined over abstract states | PCID · Diabolical combination lock | problem solved | Hurt |
| Use HOMER to learn an abstraction and explore | Diabolical combination lock, H=100 · Diabolical combination lock, H=100 | return | Helped |
Closest papers by meaning
- Representation Learning with Multi-Step Inverse Kinematics: An Efficient and Optimal Approach to Rich-Observation RL (ICML 2023 · similarity 0.92)
- Rich-Observation Reinforcement Learning with Continuous Latent Dynamics (ICML 2024 · similarity 0.89)
- Explicit Explore-Exploit Algorithms in Continuous State Spaces (NeurIPS 2019 · similarity 0.89)
- Efficient Exploration and Discriminative World Model Learning with an Object-Centric Abstraction (ICLR 2025 · similarity 0.89)
- Oracle-Efficient Reinforcement Learning for Max Value Ensembles (NeurIPS 2024 · similarity 0.88)
- Geometric Active Exploration in Markov Decision Processes: the Benefit of Abstraction (ICML 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.