Corpus ID: 210838845

Why Non-myopic Bayesian Optimization is Promising and How Far Should We Look-ahead? A Study via Rollout

@inproceedings{Yue2020WhyNB,
  title={Why Non-myopic Bayesian Optimization is Promising and How Far Should We Look-ahead? A Study via Rollout},
  author={Xubo Yue and R. Kontar},
  booktitle={AISTATS},
  year={2020}
}
  • Xubo Yue, R. Kontar
  • Published in AISTATS 2020
  • Computer Science, Mathematics
  • Lookahead, also known as non-myopic, Bayesian optimization (BO) aims to find optimal sampling policies through solving a dynamic programming (DP) formulation that maximizes a long-term reward over a rolling horizon. Though promising, lookahead BO faces the risk of error propagation through its increased dependence on a possibly mis-specified model. In this work we focus on the rollout approximation for solving the intractable DP. We first prove the improving nature of rollout in tackling… CONTINUE READING
    Can Global Optimization Strategy Outperform Myopic Strategy for Bayesian Parameter Estimation?

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 43 REFERENCES
    Practical Two-Step Lookahead Bayesian Optimization
    7
    Lookahead Bayesian Optimization with Inequality Constraints
    15
    GLASSES: Relieving The Myopia Of Bayesian Optimisation
    53
    The Knowledge Gradient Algorithm for a General Class of Online Learning Problems
    113
    Bayesian Optimization with a Finite Budget: An Approximate Dynamic Programming Approach
    37
    Multi-Task Bayesian Optimization
    348