An Asymptotically Optimal Multi-Armed Bandit Algorithm and Hyperparameter Optimization
From MaRDI portal
Publication:6344868
arXiv2007.05670MaRDI QIDQ6344868
Zhihua Zhang, Yujun Li, Hanrong Ye, Yimin Huang, Zhenguo Li
Publication date: 10 July 2020
Abstract: The evaluation of hyperparameters, neural architectures, or data augmentation policies becomes a critical model selection problem in advanced deep learning with a large hyperparameter search space. In this paper, we propose an efficient and robust bandit-based algorithm called Sub-Sampling (SS) in the scenario of hyperparameter search evaluation. It evaluates the potential of hyperparameters by the sub-samples of observations and is theoretically proved to be optimal under the criterion of cumulative regret. We further combine SS with Bayesian Optimization and develop a novel hyperparameter optimization algorithm called BOSS. Empirical studies validate our theoretical arguments of SS and demonstrate the superior performance of BOSS on a number of applications, including Neural Architecture Search (NAS), Data Augmentation (DA), Object Detection (OD), and Reinforcement Learning (RL).
Has companion code repository: https://github.com/huawei-noah/vega
This page was built for publication: An Asymptotically Optimal Multi-Armed Bandit Algorithm and Hyperparameter Optimization
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6344868)
