Research map / Graph learning / Graph algorithms and clustering
Option Discovery in the Absence of Rewards with Manifold Analysis
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 diffusion options instead of eigenoptions | Q-learning · Ring, Maze, and 4Rooms | learning convergence (average steps to goal) | Helped |
| Use diffusion options instead of eigenoptions | state-to-state exploration · Ring (t=4) | median steps between state pairs: 301 -> 217 | Helped |
| Use diffusion options instead of eigenoptions | state-to-state exploration · Maze (t=4) | median steps between state pairs: 446 -> 282 | Helped |
| Use diffusion options instead of eigenoptions | state-to-state exploration · 4Rooms (t=4) | median steps between state pairs: 160 -> 147 | Helped |
| Use diffusion options instead of cover options | state-to-state exploration · Ring, Maze, and 4Rooms | median steps between state pairs | Helped |
| Use diffusion options instead of primitive-action random walk | state-to-state exploration · Ring, Maze, and 4Rooms | median steps between state pairs | Helped |
| Increase scale from t=4 to t=13 (ablation) | Diffusion Options · Ring, Maze, and 4Rooms | learning convergence (average steps to goal) | No clear effect |
| Increase scale from t=4 to t=13 (ablation) | Diffusion Options · Maze | median steps between state pairs: 282 -> 249 | Helped |
| Increase scale from t=4 to t=13 (ablation) | Diffusion Options · 4Rooms | median steps between state pairs: 147 -> 140 | Helped |
| Increase scale from t=4 to t=13 (ablation) | Diffusion Options · Ring | median steps between state pairs: 217 -> 219 | Hurt |
| Use diffusion options instead of eigenoptions | Q-learning with downward wind · 4Rooms stochastic | learning convergence (average steps to goal) | Helped |
| Use diffusion options instead of random options | Q-learning · 4Rooms | learning convergence (average steps to goal) | Helped |
Closest papers by meaning
- Deep Laplacian-based Options for Temporally-Extended Exploration (ICML 2023 · similarity 0.91)
- A Laplacian Framework for Option Discovery in Reinforcement Learning (ICML 2017 · similarity 0.90)
- Discovering Options for Exploration by Minimizing Cover Time (ICML 2019 · similarity 0.89)
- Exploration in Reinforcement Learning with Deep Covering Options (ICLR 2020 · similarity 0.88)
- Reward-Respecting Subtasks for Model-Based Reinforcement Learning (Abstract Reprint) (AAAI 2024 · similarity 0.85)
- Beyond Optimism: Exploration With Partially Observable Rewards (NeurIPS 2024 · similarity 0.84)
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.