WBGP-BO combines multiple Gaussian processes with prefixed hyperparameters by averaging their means and standard deviations, and outperforms vanilla GP-based Bayesian optimization on several one-dimensional test problems.
Advancing Bayesian Optimization: The Mixed-Global-Local (MGL) Kernel and Length-Scale Cool Down
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
abstract
Bayesian Optimization (BO) has become a core method for solving expensive black-box optimization problems. While much research focussed on the choice of the acquisition function, we focus on online length-scale adaption and the choice of kernel function. Instead of choosing hyperparameters in view of maximum likelihood on past data, we propose to use the acquisition function to decide on hyperparameter adaptation more robustly and in view of the future optimization progress. Further, we propose a particular kernel function that includes non-stationarity and local anisotropy and thereby implicitly integrates the efficiency of local convex optimization with global Bayesian optimization. Comparisons to state-of-the art BO methods underline the efficiency of these mechanisms on global optimization benchmarks.
fields
stat.ML 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Wasserstein Barycenter Gaussian Process based Bayesian Optimization
WBGP-BO combines multiple Gaussian processes with prefixed hyperparameters by averaging their means and standard deviations, and outperforms vanilla GP-based Bayesian optimization on several one-dimensional test problems.