Statistical and Machine Learning Methods for Response Surface Optimization

Catalog description

This course presents Statistical and Machine Learning Methods for the modeling and optimization of engineering systems, for approximating and optimizing complex computer models, and for use in engineering design. Contemporary topics in Response Surface Models, including modeling systems with multiple or high dimensional responses, Probabilistic latent variables models, Optimal experimental design and its connection to Active Learning, and Manifold learning methods for Response Surface modeling. The course also treats Bayesian optimization methods based on either a physical or computer experiments using parametric (e.g., regression) models or non-parametric (e.g., Gaussian Processes or "Kriging") models, and discusses classical Response Surface methods such as Ridge Analysis and Taguchi's Robust Parameter Design.