Research map / Reinforcement learning / MDPs and planning
Reinforcement Learning with Parameterized Actions
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 |
|---|---|---|---|
| Alternate action-parameter policy updates with Q-learning for discrete action selection instead of direct eNAC policy search | Parameterized-action policy · Goal-scoring domain | goal-scoring probability: 15% -> roughly 35% | Helped |
| Learn action parameters instead of keeping them fixed at their initial values | SARSA(λ) in the goal-scoring domain · Goal-scoring domain | goal-scoring probability: roughly 20% -> roughly 35% | Helped |
| Use Q-PAMDP(1) instead of direct eNAC policy search | Parameterized-action policy · Platform domain | percentage distance covered: 10% -> 50% | Helped |
| Use Q-PAMDP(∞) instead of direct eNAC policy search | Parameterized-action policy · Platform domain | percentage distance covered: 10% -> 65% | Helped |
| Use Q-PAMDP(∞) instead of Q-PAMDP(1), performing full parameter optimization before relearning the discrete-action values (ablation) | Q-PAMDP · Platform domain | percentage distance covered: 50% -> 65% | Helped |
| Use Q-PAMDP(∞) instead of Q-PAMDP(1) (ablation) | Q-PAMDP · Goal-scoring domain | goal-scoring probability: roughly 35% -> roughly 35% | No clear effect |
| Use Q-PAMDP(1) instead of fixed-parameter SARSA | Parameterized-action policy · Platform domain | percentage distance covered: 40% -> 50% | Helped |
| Use Q-PAMDP(∞) instead of fixed-parameter SARSA | Parameterized-action policy · Platform domain | percentage distance covered: 40% -> 65% | Helped |
Closest papers by meaning
- Model-based Reinforcement Learning for Parameterized Action Spaces (ICML 2024 · similarity 0.91)
- Context-Sensitive Abstractions for Reinforcement Learning with Parameterized Actions (AAAI 2026 · similarity 0.90)
- Meta-learning Parameterized Skills (ICML 2023 · similarity 0.88)
- Learning to Act in Decentralized Partially Observable MDPs (ICML 2018 · similarity 0.87)
- How RL Agents Behave When Their Actions Are Modified (AAAI 2021 · similarity 0.87)
- Scalable Policy-Based RL Algorithms for POMDPs (NeurIPS 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.