Optimization and Machine Learning

Abstract

The course teaches the basics of nonlinear optimization and concepts of machine learning. An introduction to the finite element method allows an extension of the application area to real engineering problems such as structural optimization and modeling of material behavior on different length scales.

Objective

Students will learn mathematical optimization methods including gradient based and gradient free methods as well as established algorithms in the context of machine learning to solve real engineering problems, which are generally non-linear in nature. Strategies to ensure efficient training of machine learning models based on large data sets define another teaching goal of the course. Optimization tools (MATLAB, LS-Opt, Python) and the finite element program ABAQUS are presented to solve both general and real engineering problems.

Content

- Introduction into Nonlinear Optimization - Design of Experiments DoE - Introduction into Nonlinear Finite Element Analysis - Optimization based on Meta Modeling Techniques - Shape and Topology Optimization - Robustness and Sensitivity Analysis - Fundamentals of Machine Learning - Generalized methods for regression and classification, Neural Networks, Support Vector machines - Supervised and unsupervised learning

Resources