Bayesian Optimization: Exploiting Machine Learning Models, Physics, and Throughput Experiments

Explore new paradigms for Bayesian Optimization (BO) in this 1-hour 5-minute webinar presented by Victor M. Zavala, Baldovin-DaPra Professor at the University of Wisconsin-Madison. Learn about innovative approaches that leverage large-scale machine learning models, physical knowledge, and high-throughput experiments to enhance optimization processes. Discover a method that decomposes performance functions into reference and residual models, accelerating searches through Gaussian Process learning. Understand how reference models can be used to partition design spaces and enable parallel searches for high-throughput experiments. Examine a BO implementation that incorporates large-scale, parametric models with scalable uncertainty quantification capabilities. Gain insights from real-world applications in controller tuning for energy systems, reactor optimization, and microbial community design. This webinar, part of the Data-Driven Physical Simulations series, offers valuable knowledge for researchers and practitioners in fields such as chemical engineering, computational mathematics, and data-driven optimization.

DDPS | Bayesian Optimization: Exploiting Machine Learning Models, Physics, & Throughput Experiments