Corpus ID: 62841559

Fast Efficient Hyperparameter Tuning for Policy Gradients

@article{Paul2019FastEH,
  title={Fast Efficient Hyperparameter Tuning for Policy Gradients},
  author={Supratik Paul and Vitaly Kurin and Shimon Whiteson},
  journal={ArXiv},
  year={2019},
  volume={abs/1902.06583}
}
  • Supratik Paul, Vitaly Kurin, Shimon Whiteson
  • Published 2019
  • Mathematics, Computer Science
  • ArXiv
  • The performance of policy gradient methods is sensitive to hyperparameter settings that must be tuned for any new application. Widely used grid search methods for tuning hyperparameters are sample inefficient and computationally expensive. More advanced methods like Population Based Training that learn optimal schedules for hyperparameters instead of fixed settings can yield better results, but are also sample inefficient and computationally expensive. In this paper, we propose Hyperparameter… CONTINUE READING

    Figures, Tables, and Topics from this paper.

    One-Shot Bayes Opt with Probabilistic Population Based Training
    Smoothing Policies and Safe Policy Gradients
    6
    AReN: assured ReLU NN architecture for model predictive control of LTI systems
    2
    Online Hyper-parameter Tuning in Off-policy Learning via Evolutionary Strategies

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 54 REFERENCES
    Scalable Gradient-Based Tuning of Continuous Regularization Hyperparameters
    58
    Population Based Training of Neural Networks
    211
    Hyperparameter optimization with approximate gradient
    102
    Where Did My Optimum Go?: An Empirical Analysis of Gradient Descent Optimization in Policy Gradient Methods
    13
    Fast Bayesian Optimization of Machine Learning Hyperparameters on Large Datasets
    190
    Practical Bayesian Optimization of Machine Learning Algorithms
    3156
    Multi-Task Bayesian Optimization
    345
    Gradient-Based Optimization of Hyperparameters
    261
    A Greedy Approach to Adapting the Trace Parameter for Temporal Difference Learning
    18