Stanford AA222/CS361 Engineering Design Optimization I Probabilistic Surrogate Optimization

In this lecture for Stanford's AA 222 / CS 361 Engineering Design Optimization course, we dive into the intricacies of Probabilistic Surrogate Optimization. The content covers key methodologies, including the development and use of surrogate models for efficient optimization of complex engineering designs. These comprehensive models are presented as critical tools for the evaluation and improvement of design performances. The lecture also emphasizes the application of probabilistic methods for managing uncertainty and improving decision-making in the design process. Lecture Outline Surrogate Model Selection Probabilistic Surrogate Models Gaussian Distributions Gaussian Processes Prediction Noisy Measurements Fitting Gaussian Processes Surrogate Optimization Exploration Prediction-based Error-based Lower Confidence Bound Probability of Improvement Expected Improvement Notebook: https://github.com/josh0tt/SurrogateOptimization View the course website: https://aa222.stanford.edu/