Research map / Reinforcement learning / Visual and representation RL
Flexible Option 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 |
|---|---|---|---|
| Update all relevant options instead of only the sampled option | Option-critic in FourRooms with fixed hallway options · FourRooms | steps per episode | Helped |
| Update all relevant options | Option-critic, compared with OC · FourRooms | episodes to reach the same performance | Helped |
| Update all relevant options | Option-critic, compared with OC · MountainCar transfer | performance and sample efficiency | Helped |
| Increase the probability of updating all options (η) (ablation) | MOC · MountainCar transfer | final performance | Helped |
| Update all options instead of only the sampled option | Option-critic, compared with IOC and PPO · MuJoCo transfer tasks | performance and sample efficiency | Helped |
| Use eight options rather than fewer options (ablation) | MOC · Walker2dStand-v1 | performance: 617 (98) -> 535 (81) | Hurt |
| Use eight options rather than fewer options (ablation) | IOC · Walker2dStand-v1 | performance: 326 (92) -> 212 (64) | Hurt |
| Set η=1.0 rather than using intermediate η (ablation) | MOC · FourRooms | option diversity (information radius) | Hurt |
| Use intermediate η values instead of η=1.0 (ablation) | MOC · FourRooms | option diversity (information radius) | Helped |
| Increase η (ablation) | MOC · MountainCar | option diversity (information radius) | No clear effect |
| Use MOC rather than IOC | Option-critic with learned option policies · MuJoCo option duration | option duration | Helped |
| Use MOC rather than IOC | PPO-based hierarchical agents · MuJoCo with independent second-to-last hidden layers | performance robustness | Helped |
Closest papers by meaning
- The Option-Critic Architecture (AAAI 2017 · similarity 0.93)
- Autonomous Option Invention for Continual Hierarchical Reinforcement Learning and Planning (AAAI 2025 · similarity 0.92)
- Reward-Respecting Subtasks for Model-Based Reinforcement Learning (Abstract Reprint) (AAAI 2024 · similarity 0.90)
- Option-aware Temporally Abstracted Value for Offline Goal-Conditioned Reinforcement Learning (NeurIPS 2025 · similarity 0.89)
- Rethinking Decision Transformer via Hierarchical Reinforcement Learning (ICML 2024 · similarity 0.88)
- Pausing Policy Learning in Non-stationary Reinforcement Learning (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.