2019/10/03 by David Eriksson, Michael Pearce, Jacob R Gardner +2 · 2 voices · 5 citations
Computer Science · Mathematics · #cs.LG #stat.ML
published as In Advances in Neural Information Processing Systems 32, pages 5497-5508. 2019 · Appears in NeurIPS 2019 as a spotlight paper
arxiv published 2019/10/03 · arxiv created 2020/02/24 · arxiv updated 2020/02/26
Bayesian optimization has recently emerged as a popular method for the sample-efficient optimization of expensive black-box functions. However, the application to high-dimensional problems with several thousand observations remains challenging, and on difficult problems Bayesian optimization is often not competitive with other paradigms. In this paper we take the view that this is due to the implicit homogeneity of the global probabilistic models and an overemphasized exploration that results from global acquisition. This motivates the design of a local probabilistic approach for global optimization of large-scale high-dimensional problems. We propose the TuRBO algorithm that fits a collection of local models and performs a principled global allocation of samples across these models via an implicit bandit approach. A comprehensive evaluation demonstrates that TuRBO outperforms state-of-the-art methods from machine learning and operations research on problems spanning reinforcement learning, robotics, and the natural sciences.