Research map / Reinforcement learning / Visual and representation RL
Effectively Learning Initiation Sets in Hierarchical Reinforcement Learning
Akhil Bagaria, Ben M Abbatematteo, Omer Gottesman, Matt Corsaro, Sreehari Rammohan, George Konidaris
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| learn initiation sets with IVF or Weighted classifier instead of binary classifier | option initiation set learner · MiniGrid-FourRooms and Montezuma's Revenge | initiation set accuracy and size | Improved |
| use binary classification over training time (baseline behavior) | baseline binary classifier · FourRooms and Montezuma's Revenge | initiation set accuracy over time | Hurt |
| add optimistic exploration bonus to IVF (ablation) | IVF · Montezuma's Revenge | initiation set quality | Improved |
| add optimistic exploration bonus to IVF (ablation) | IVF · MiniGrid-FourRooms | initiation set quality | Hurt |
| remove the optimistic count-based bonus (ablation) | Weighted classifier · Robosuite SLIDE | task success rate | Hurt |
| use Weighted instead of Random/Binary | initiation learner · Robosuite DOOR | task success rate | No clear effect |
| use Weighted instead of IVF/Binary/Random | initiation learner · Robosuite LEVER and SLIDE | task success rate: above 50% | Improved |
| add IVF or Weighted initiation learning to deep skill chaining | DSC · Ant Medium-Maze | success rate | Improved |
| use Weighted classifier instead of pure IVF | DSC · Ant Medium-Maze | mean return and variance | Mixed |
Closest papers by meaning
- Jump-Start Reinforcement Learning (ICML 2023 · similarity 0.88)
- GO-DICE: Goal-Conditioned Option-Aware Offline Imitation Learning via Stationary Distribution Correction Estimation (AAAI 2024 · similarity 0.87)
- Autonomous Option Invention for Continual Hierarchical Reinforcement Learning and Planning (AAAI 2025 · similarity 0.87)
- Option-aware Temporally Abstracted Value for Offline Goal-Conditioned Reinforcement Learning (NeurIPS 2025 · similarity 0.86)
- Reinforcement Learning with Lookahead Information (NeurIPS 2024 · similarity 0.86)
- Reward-Respecting Subtasks for Model-Based Reinforcement Learning (Abstract Reprint) (AAAI 2024 · 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.
