Corpus ID: 14542167

MazeBase: A Sandbox for Learning from Games

@article{Sukhbaatar2015MazeBaseAS,
  title={MazeBase: A Sandbox for Learning from Games},
  author={Sainbayar Sukhbaatar and Arthur Szlam and Gabriel Synnaeve and Soumith Chintala and R. Fergus},
  journal={ArXiv},
  year={2015},
  volume={abs/1511.07401}
}
  • Sainbayar Sukhbaatar, Arthur Szlam, +2 authors R. Fergus
  • Published 2015
  • Computer Science
  • ArXiv
  • This paper introduces MazeBase: an environment for simple 2D games, designed as a sandbox for machine learning approaches to reasoning and planning. Within it, we create 10 simple games embodying a range of algorithmic tasks (e.g. if-then statements or set negation). A variety of neural models (fully connected, convolutional network, memory network) are deployed via reinforcement learning on these games, with and without a procedurally generated curriculum. Despite the tasks' simplicity, the… CONTINUE READING
    71 Citations

    Figures, Tables, and Topics from this paper

    TextWorld: A Learning Environment for Text-based Games
    • 82
    • PDF
    APES: a Python toolbox for simulating reinforcement learning environments
    • 1
    • PDF
    ELF: An Extensive, Lightweight and Flexible Research Platform for Real-time Strategy Games
    • 95
    • Highly Influenced
    • PDF
    Value Propagation Networks
    • 11
    • PDF
    Control of Memory, Active Perception, and Action in Minecraft
    • 188
    • Highly Influenced
    • PDF
    BABYAI: FIRST STEPS TOWARDS GROUNDED LAN-
    Interactive Grounded Language Acquisition and Generalization in a 2D World
    • 45
    • PDF
    CIENCY OF GROUNDED LANGUAGE LEARNING
    • Highly Influenced
    Deep Successor Reinforcement Learning For Maze Navigation
    • 1
    • PDF

    References

    SHOWING 1-10 OF 35 REFERENCES
    The Arcade Learning Environment: An Evaluation Platform for General Agents (Extended Abstract)
    • 1,459
    • Highly Influential
    • PDF
    Fast Heuristic Search for RTS Game Combat Scenarios
    • 118
    • PDF
    A Bayesian model for RTS units control applied to StarCraft
    • 60
    • PDF
    Applying reinforcement learning to small scale combat in the real-time strategy game StarCraft:Broodwar
    • S. Wender, I. Watson
    • Computer Science
    • 2012 IEEE Conference on Computational Intelligence and Games (CIG)
    • 2012
    • 96
    • PDF
    A Survey of Real-Time Strategy Game AI Research and Competition in StarCraft
    • 345
    • PDF
    Deep Learning for Real-Time Atari Game Play Using Offline Monte-Carlo Tree Search Planning
    • 272
    • Highly Influential
    • PDF
    Learning Simple Algorithms from Examples
    • 87
    • PDF
    Inferring Algorithmic Patterns with Stack-Augmented Recurrent Nets
    • 297
    • Highly Influential
    • PDF
    Human-level control through deep reinforcement learning
    • 10,966
    • PDF
    Playing Atari with Deep Reinforcement Learning
    • 4,584
    • Highly Influential
    • PDF