Triangulation candidates for Bayesian optimization

2021/12/14 by Gramacy, Robert B., Sauer, Annie, Wycoff, Nathan · 3 citations
#Computation (stat.CO) #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML)

paper · doi:10.48550/arxiv.2112.07457

Abstract

Bayesian optimization involves "inner optimization" over a new-data acquisition criterion which is non-convex/highly multi-modal, may be non-differentiable, or may otherwise thwart local numerical optimizers. In such cases it is common to replace continuous search with a discrete one over random candidates. Here we propose using candidates based on a Delaunay triangulation of the existing input design. We detail the construction of these "tricands" and demonstrate empirically how they outperform both numerically optimized acquisitions and random candidate-based alternatives, and are well-suited for hybrid schemes, on benchmark synthetic and real simulation experiments.

Cited by

• Profile Bayesian Optimization for Expensive Computer Experiments
• Actively Learning Joint Contours of Multiple Computer Experiments
• Staying on Track: Efficient Trajectory Discovery with Adaptive Batch Sampling

Related

• Composition of kernel and acquisition functions for High Dimensional Bayesian Optimization
• Voronoi Candidates for Bayesian Optimization
• Information-theoretic Bayesian Optimization: Survey and Tutorial
• Efficient Gradient-Enhanced Bayesian Optimizer with Comparisons to Conjugate-Gradient and Quasi-Newton Optimizers for Unconstrained Local Optimization
• High Dimensional Bayesian Optimization Using Dropout