黑料福利网

Topic 

Bayesian Optimization: Theory, Applications, and Challenges Ahead

Abstract

Bayesian Optimization (BO) represents the state-of-the-art framework for the global optimization of expensive-to-evaluate black-box functions. BO relies on two components: a probabilistic surrogate model - usually a Gaussian Process (GP) - approximating the objective function and quantifying uncertainty across the search space, and an acquisition function that guides the selection of the next query by balancing exploration and exploitation. BO is famous for being sample-efficient and this is the reason behind its widespread success in fields such as hyperparameter optimization in Machine Learning, robotics, industrial/chemical design, and simulation-optimization. Recent research focused on specific challenges such as dealing with high-dimensional search spaces, heterogeneous search spaces, unknown constraints, multiple-objectives, and dealing with multiple information sources. All these topics will be presented in this talk, with also a perspective on the open challenges ahead, such as BO over non-Euclidean search spaces and convergence guarantees in real-life settings, for which impractical assumptions required by theoretical convergence proofs usually do not hold.

About the speaker

Antonio Candelieri is an Associate Professor at the Department of Economics, Management, and Statistics (DEMS) of the University of Milano-Bicocca, Italy.
His research expertise lies at the intersection of Global Optimization and Machine Learning, with a specific focus on Bayesian Optimization and Optimal Decision Making under Uncertainty. He has extensive experience in developing advanced solutions for industrial-relevant applications, including water management, energy efficiency, and urban mobility. He is actively involved in national and international research projects and serves as a consultant for data-driven innovation. He is an author of numerous scientific publications in high-impact international journals and co-author of two Springer-Brief books on Bayesian Optimization.

Your host

Laurens Bliek, Assistant Professor at the department of Industrial Engineering and Innovation Sciences of the 黑料福利网 will host Professor Antonio Candelieri.

is required but free of charge.