Research map / Reinforcement learning / Visual and representation RL
A Laplacian Framework for Option Discovery in 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 |
|---|---|---|---|
| Add enough eigenoptions to the behavior policy | Agents using primitive actions · 10×10 grid, I-Maze, and 4-room domain | diffusion time | Helped |
| Add only a few eigenoptions to the behavior policy | Agents using primitive actions · 10×10 grid, I-Maze, and 4-room domain | diffusion time | Hurt |
| Use only options that lead to bottleneck states | Random exploration in the 4-room domain · 4-room domain | diffusion time | Hurt |
| Add eigenoptions to the behavior policy | Q-learning agent · 10×10 grid, I-Maze, and 4-room domain | final performance | Helped |
| Use only a few eigenoptions | Q-learning agent · 10×10 grid, I-Maze, and 4-room domain | final performance | Hurt |
| Use 64 eigenoptions | Q-learning agent · 10×10 grid, I-Maze, and 4-room domain | final performance | Helped |
| Add eigenoptions to the behavior policy | Q-learning agent · 4-room domain, multiple start and goal states | final performance | Helped |
| Use eigenoptions instead of doorway-seeking bottleneck options | Q-learning agent · 4-room domain, goals far from doorways | final performance | Helped |
| Use doorway-seeking bottleneck options instead of eigenoptions | Q-learning agent · 4-room domain, goals near bottleneck states | final performance | Mixed |
| Replace eigenoptions with random-subgoal options | Random exploration policy · 4-room domain | diffusion time | Hurt |
| Replace eigenoptions with random-subgoal options | Q-learning agent · 4-room domain | learning performance | Hurt |
| Use eigenoption #445, discovered without the game's reward signal | Atari agent · Freeway | score: 28 points | Helped |
Closest papers by meaning
- Eigenoption Discovery through the Deep Successor Representation (ICLR 2018 · similarity 0.91)
- Option Discovery in the Absence of Rewards with Manifold Analysis (ICML 2020 · similarity 0.90)
- Reward-Respecting Subtasks for Model-Based Reinforcement Learning (Abstract Reprint) (AAAI 2024 · similarity 0.87)
- Autonomous Option Invention for Continual Hierarchical Reinforcement Learning and Planning (AAAI 2025 · similarity 0.86)
- Reward-Aware Proto-Representations in Reinforcement Learning (NeurIPS 2025 · similarity 0.86)
- Online Laplacian-Based Representation Learning in Reinforcement Learning (ICML 2025 · similarity 0.86)
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.