Research map / Reinforcement learning / Visual and representation RL
SDRL: Interpretable and Data-Efficient Deep Reinforcement Learning Leveraging Symbolic Planning
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 symbolic-transition subtasks with intrinsic-goal planning and learned controllers | SDRL · Montezuma's Revenge | samples to reach cumulative external reward 400: around 1.5M samples; hDQN cannot reliably achieve the score of 400 around 2.5M samples | Helped |
| Use intrinsic goals that can change with the reward structure rather than a fixed explicit planning goal | SDRL · Taxi domain | policy adaptation to the optimal reward-seeking behavior | Helped |
| Define subtasks as symbolic transitions rather than object-associated goals | SDRL · Montezuma's Revenge | learning stability | Helped |
| Use symbolic-transition subtasks and intrinsic-goal planning instead of the fixed goal in standard PEORL | SDRL · Taxi domain | initial convergence speed | Hurt |
| Use planning-based intrinsic-goal exploration rather than model-free exploration | SDRL · Taxi domain, Task 1 | initial convergence speed | Hurt |
Closest papers by meaning
- BlendRL: A Framework for Merging Symbolic and Neural Policy Learning (ICLR 2025 · similarity 0.88)
- Mitigating Information Loss in Tree-Based Reinforcement Learning via Direct Optimization (ICLR 2025 · similarity 0.88)
- Contextual Pre-planning on Reward Machine Abstractions for Enhanced Transfer in Deep Reinforcement Learning (AAAI 2024 · similarity 0.87)
- End-to-End Neuro-Symbolic Reinforcement Learning with Textual Explanations (ICML 2024 · similarity 0.87)
- Autonomous Option Invention for Continual Hierarchical Reinforcement Learning and Planning (AAAI 2025 · similarity 0.86)
- Representation Learning Enables Scalable Multitask Deep Reinforcement Learning (NeurIPS 2026 · 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.