Corpus ID: 67856569

Unifying Ensemble Methods for Q-learning via Social Choice Theory

@article{Chourasia2019UnifyingEM,
  title={Unifying Ensemble Methods for Q-learning via Social Choice Theory},
  author={R. Chourasia and A. Singla},
  journal={ArXiv},
  year={2019},
  volume={abs/1902.10646}
}
  • R. Chourasia, A. Singla
  • Published 2019
  • Computer Science
  • ArXiv
  • Ensemble methods have been widely applied in Reinforcement Learning (RL) in order to enhance stability, increase convergence speed, and improve exploration. [...] Key Method We map the problem of designing an action aggregation mechanism in an ensemble method to a voting problem which, under different voting rules, yield popular ensemble-based RL algorithms like Majority Voting Q-learning or Bootstrapped Q-learning.Expand Abstract

    Figures, Tables, and Topics from this paper.

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 46 REFERENCES
    Ensemble Algorithms in Reinforcement Learning
    108
    Budgeted Social Choice: From Consensus to Personalized Decision Making
    171
    Neural Network Ensembles in Reinforcement Learning
    29
    Multi-step Reinforcement Learning: A Unifying Algorithm
    49
    Properties of multiwinner voting rules
    155
    Justified Representation in Multiwinner Voting: Axioms and Algorithms
    2
    Ensemble Machine Learning: Methods and Applications
    270
    (More) Efficient Reinforcement Learning via Posterior Sampling
    225
    Representative Deliberations and Representative Decisions: Proportional Representation and the Borda Rule
    233
    The Condorcet Principle for Multiwinner Elections: From Shortlisting to Proportionality
    31