[Colloquium] Fengxue Zhang MS Presentation/Apr 11, 2023
Megan Woodward
meganwoodward at uchicago.edu
Mon Mar 27 15:24:35 CDT 2023
This is an announcement of Fengxue Zhang's MS Presentation
===============================================
Candidate: Fengxue Zhang
Date: Tuesday, April 11, 2023
Time: 12:30 pm CST
Location: JCL 298
M.S. Paper Title: Learning Regions of Interest for Bayesian Optimization with Adaptive Level-Set Estimation
Abstract: We study Bayesian optimization (BO) in high-dimensional and non-stationary scenarios. Existing algorithms for such scenarios typically require extensive hyperparameter tuning, which limits their practical effectiveness. We propose a framework, called BALLET, which adaptively filters for a high-confidence region of interest (ROI) as a superlevel-set of a nonparametric probabilistic model such as a Gaussian process (GP). Our approach is easy to tune, and is able to focus on local region of the optimization space that can be tackled by existing BO methods. The key idea is to use two probabilistic models: a coarse GP to identify the ROI, and a localized GP for optimization within the ROI. We show theoretically that BALLET can efficiently shrink the search space, and can exhibit a tighter regret bound than standard BO without ROI filtering. We demonstrate empirically the effectiveness of BALLET on both synthetic and real-world optimization tasks.
Advisors: Yuxin Chen
Committee Members: Rebecca Willett, Haifeng Xu, and Yuxin Chen
-------------- next part --------------
A non-text attachment was scrubbed...
Name: BALLET_Qual23.pdf
Type: application/pdf
Size: 669755 bytes
Desc: BALLET_Qual23.pdf
URL: <http://mailman.cs.uchicago.edu/pipermail/colloquium/attachments/20230327/b426a954/attachment-0001.pdf>
More information about the Colloquium
mailing list