In this course, we will dive into the details of the most recent applications in data driven engineering within the scope of machine learning (ML). Building upon the skills developed in the “Data Driven Engineering 1: Machine Learning for Dynamical Systems” course, students will learn about complex model architectures through different “themes”, with the objective of providing a deeper background and capability to navigate through the recent developments in the field.

The lecture is integrated with group projects. Students will form groups and be assigned open-ended research problems with scientific mentors at the beginning of the semester. In the following weeks, the groups will work on their problems by combining the fundamental skills they gained in the first and second lectures. The progress will be monitored via project sessions throughout the semester. The finalized work will be presented at the end of the semester and published in the lecture repository.

Contents:

1. Introduction to the lecture and project workflow
2. Data-driven image processing for fluid mechanics
3. Dynamic Mode Decomposition and coordinate transformations
4. Modelling of transport phenomena with neural networks
5. State space models
6. Integration of genetic algorithms with machine learning
7. ML-Augmented Experiment Design and Machine Learning Control
8. Project presentations

Lecture Format:

• Lectures: 45 min; Practice hours: 45 minutes
• Project sessions: 90 minutes

Workload:

• Regular attendance: 21 hours
• Self-study: 42 hours

Learning Objectives:

Students have the ability to:

• explain the underlying mathematics of ML algorithms,
• tailor data-driven methods on open-ended engineering problems,
• tackle challenges for complex, hybrid architectures,
• describe and apply various architectures for specialized domains such as flow visualization, transport phenomena, experiment design,
• extract patterns and correlate them with the physics of the problem for dynamical datasets,
• navigate through the recent developments in the field,
• plan and conduct research projects in groups.

Recommendations:

The course is intended for students with a strong background and interest in ML applications for engineering problems. It is strongly recommended to be taken in combination with the “Data Driven Engineering 1 Machine Learning for Dynamical Systems” course.