Potential-Based Shaping and Q-Value Initialization are Equivalent

@article{Wiewiora2003PotentialBasedSA,
  title={Potential-Based Shaping and Q-Value Initialization are Equivalent},
  author={Eric Wiewiora},
  journal={ArXiv},
  year={2003},
  volume={abs/1106.5267}
}
  • Eric Wiewiora
  • Published in J. Artif. Intell. Res. 2003
  • Computer Science
  • Shaping has proven to be a powerful but precarious means of improving reinforcement learning performance. Ng, Harada, and Russell (1999) proposed the potential-based shaping algorithm for adding shaping rewards in a way that guarantees the learner will learn optimal behavior. In this note, we prove certain similarities between this shaping algorithm and the initialization step required for several reinforcement learning algorithms. More specifically, we prove that a reinforcement learner… CONTINUE READING

    Create an AI-powered research feed to stay up to date with new papers like this posted to ArXiv

    Citations

    Publications citing this paper.
    SHOWING 1-10 OF 89 CITATIONS

    Theoretical considerations of potential-based reward shaping for multi-agent systems

    VIEW 12 EXCERPTS
    CITES BACKGROUND & RESULTS
    HIGHLY INFLUENCED

    Learning Traffic Signal Control with Advice

    VIEW 4 EXCERPTS
    CITES BACKGROUND
    HIGHLY INFLUENCED

    Distributed reinforcement learning for network intrusion response

    VIEW 4 EXCERPTS
    CITES BACKGROUND
    HIGHLY INFLUENCED

    Dynamic potential-based reward shaping

    VIEW 6 EXCERPTS
    CITES BACKGROUND & METHODS
    HIGHLY INFLUENCED

    Policy Invariance under Reward Transformations for General-Sum Stochastic Games

    VIEW 3 EXCERPTS
    CITES BACKGROUND & METHODS
    HIGHLY INFLUENCED

    An Emotion-Based Approach to Reinforcement Learning Reward Design

    • Haixu Yu, Pei Yang
    • Computer Science
    • 2019 IEEE 16th International Conference on Networking, Sensing and Control (ICNSC)
    • 2019
    VIEW 1 EXCERPT
    CITES BACKGROUND

    An modeling processing method for video games based on deep reinforcement learning

    VIEW 1 EXCERPT
    CITES BACKGROUND

    FILTER CITATIONS BY YEAR

    2005
    2019

    CITATION STATISTICS

    • 5 Highly Influenced Citations

    • Averaged 8 Citations per year from 2017 through 2019

    • 20% Increase in citations per year in 2019 over 2018

    References

    Publications referenced by this paper.
    SHOWING 1-2 OF 2 REFERENCES

    Poli y invarian e under reward transformations : theory and appli ation to reward shaping

    • A. G. Barto
    • Ma hine Learning , Pro eedings of theSixteenth International Conferen e
    • 1999

    The e e t of representation and knowledge on goal - dire ted exploration with reinfor ement - learning algorithms

    • A. Y. Ng, D. Harada, S. Russell
    • Ma hine Learning
    • 1996