Corpus ID: 14320659

Global Policy Construction in Modular Reinforcement Learning

@inproceedings{Zhang2015GlobalPC,
  title={Global Policy Construction in Modular Reinforcement Learning},
  author={Ruohan Zhang and Z. Song and D. Ballard},
  booktitle={AAAI},
  year={2015}
}
  • Ruohan Zhang, Z. Song, D. Ballard
  • Published in AAAI 2015
  • Computer Science
  • We propose a modular reinforcement learning algorithm which decomposes a Markov decision process into independent modules. Each module is trained using Sarsa(λ). We introduce three algorithms for forming global policy from modules policies, and demonstrate our results using a 2D grid world. 
    3 Citations

    Figures and Topics from this paper

    Composable Modular Reinforcement Learning
    • 3
    • PDF
    Modeling sensory-motor decisions in natural behavior
    • 5
    • PDF
    Modeling sensory-motor decisions in natural behavior

    References

    SHOWING 1-7 OF 7 REFERENCES
    Multiple-Goal Reinforcement Learning with Modular Sarsa(0)
    • 86
    • PDF
    Introduction to Reinforcement Learning
    • 5,405
    Q-Decomposition for Reinforcement Learning Agents
    • 125
    • PDF
    The MAXQ Method for Hierarchical Reinforcement Learning
    • 318
    • PDF
    Multiple Model-Based Reinforcement Learning
    • 423
    • PDF
    Generalization in Reinforcement Learning: Safely Approximating the Value Function
    • 635
    • PDF
    Action Selection methods using Reinforcement Learning
    • 83
    • Highly Influential