Bayesian Optimization for Discrete, Missing and Cost-sensitive Inputs
thesis
posted on 2023-08-11, 03:12authored byHuu Phuc Luong
This thesis addresses three important problems in practical Bayesian optimization, an important area of study in computer science: discrete inputs, missing values, and cost-sensitive inputs. Two novel approaches are proposed to allow the Bayesian optimization approaches to be more applicable and successful in many practical situations where there are discrete or missing input values. In addition, a framework is proposed to handle a given optimization budget in an elegant way that automatically selects a suitable searching strategy. These findings are significant for researchers working in the field of Bayesian optimization and inspire future research.