Corpus ID: 13528549

Horde: a scalable real-time architecture for learning knowledge from unsupervised sensorimotor interaction

  title={Horde: a scalable real-time architecture for learning knowledge from unsupervised sensorimotor interaction},
  author={R. Sutton and Joseph Modayil and M. Delp and T. Degris and P. Pilarski and Adam White and Doina Precup},
  • R. Sutton, Joseph Modayil, +4 authors Doina Precup
  • Published in AAMAS 2011
  • Computer Science
  • Maintaining accurate world knowledge in a complex and changing environment is a perennial problem for robots and other artificial intelligence systems. Our architecture for addressing this problem, called Horde, consists of a large number of independent reinforcement learning sub-agents, or demons. Each demon is responsible for answering a single predictive or goal-oriented question about the world, thereby contributing in a factored, modular way to the system's overall knowledge. The questions… CONTINUE READING
    321 Citations
    Composable Action-Conditioned Predictors: Flexible Off-Policy Learning for Robot Navigation
    • 9
    • Highly Influenced
    • PDF
    Visual Reinforcement Learning with Imagined Goals
    • 180
    • PDF
    BECCA: Reintegrating AI for Natural World Interaction
    • B. Rohrer
    • Computer Science
    • AAAI Spring Symposium: Designing Intelligent Robots
    • 2012
    • 12
    Self-organizing maps for storage and transfer of knowledge in reinforcement learning
    • 10
    • PDF
    Multi-timescale nexting in a reinforcement learning robot
    • 94
    • PDF
    Accelerating Learning in Constructive Predictive Frameworks with the Successor Representation
    • 5
    • PDF


    Map Learning with Uninterpreted Sensors and Effectors
    • 217
    • PDF
    A Method for Clustering the Experiences of a Mobile Robot that Accords with Human Judgments
    • 96
    • PDF
    Integrated Architectures for Learning, Planning, and Reacting Based on Approximating Dynamic Programming
    • 1,432
    • PDF
    Linking Action to Perception in a Humanoid Robot: a Developmental Approach to Grasping
    • 46
    • PDF
    Reinforcement Learning: An Introduction
    • 27,059
    • PDF
    Learning in Worlds with Objects
    • 20
    • PDF
    Neo: learning conceptual knowledge by sensorimotor interaction with an environment
    • 49
    • PDF
    GQ(λ): A general gradient algorithm for temporal-difference prediction learning with eligibility traces
    • 94
    • PDF