Corpus ID: 12537077

Compress and Control

@inproceedings{Veness2015CompressAC,
  title={Compress and Control},
  author={J. Veness and Marc G. Bellemare and Marcus Hutter and Alvin Chua and G. Desjardins},
  booktitle={AAAI},
  year={2015}
}
  • J. Veness, Marc G. Bellemare, +2 authors G. Desjardins
  • Published in AAAI 2015
  • Computer Science, Mathematics
  • This paper describes a new information-theoretic policy evaluation technique for reinforcement learning. This technique converts any compression or density model into a corresponding estimate of value. Under appropriate stationarity and ergodicity conditions, we show that the use of a sufficiently powerful model gives rise to a consistent value function estimator. We also study the behavior of this technique when applied to various Atari 2600 video games, where the use of suboptimal modeling… CONTINUE READING
    20 Citations
    Increasing the Action Gap: New Operators for Reinforcement Learning
    • 88
    • PDF
    A Distributional Perspective on Reinforcement Learning
    • 465
    • PDF
    State Abstraction as Compression in Apprenticeship Learning
    • 20
    • PDF
    Nonparametric General Reinforcement Learning
    • 12
    • PDF
    TD or not TD: Analyzing the Role of Temporal Differencing in Deep Reinforcement Learning
    • 11
    • PDF
    Unifying Count-Based Exploration and Intrinsic Motivation
    • 619
    • PDF
    A Deep Learning Approach for Joint Video Frame and Reward Prediction in Atari Games
    • 30
    • Highly Influenced
    • PDF
    Efficient Exploration in Reinforcement Learning through Time-Based Representations
    • 1
    • PDF
    Counterfactual Credit Assignment in Model-Free Reinforcement Learning
    • PDF

    References

    SHOWING 1-10 OF 52 REFERENCES
    Dual Representations for Dynamic Programming and Reinforcement Learning
    • Tao Wang, Michael Bowling, Dale Schuurmans
    • Computer Science
    • 2007 IEEE International Symposium on Approximate Dynamic Programming and Reinforcement Learning
    • 2007
    • 28
    • PDF
    Reinforcement Learning via AIXI Approximation
    • 26
    • PDF
    A Monte-Carlo AIXI Approximation
    • 149
    • PDF
    Cover tree Bayesian reinforcement learning
    • 17
    • PDF
    Modelling Sparse Dynamical Systems with Compressed Predictive State Representations
    • 39
    • PDF
    Adaptive Subgradient Methods for Online Learning and Stochastic Optimization
    • 6,425
    • PDF
    Integrating Sample-Based Planning and Model-Based Reinforcement Learning
    • 106
    • PDF
    Efficient Bayes-Adaptive Reinforcement Learning using Sample-Based Search
    • 118
    • PDF
    Algorithms for Reinforcement Learning
    • 844
    • Highly Influential
    • PDF
    Stable Dual Dynamic Programming
    • 26
    • PDF