Tuesday, 21 November 2023
 

Time:
3:00 - 4:30 PM
Place: This Tutorial is a hybrid event. You can join the event on-site and online.
Material Engineering Center Saarland (MECS), Saarland University
Campus, Building D3.3, Room 2.15 (1. Floor, seminar room)
Moderator:
Dr. Tim Dahmen, German Research Center for Artificial Intelligence (DFKI)

Deep Learning (DL) has tremendous impact in various scientific fields including Materials Science and Engineering (MSE) and has very successfully been used for various tasks in the field. However, the method is both compute- and data-hungry. In this tutorial, we discuss options to generate training data synthetically. The advantages are numerous: (1) one create arbitrary amounts of data, (2) one can precisely control class balance according to training needs, (3) data annotations are trivially included, and (4) the data can contain any thinkable cases (instead of only observed cases).

However, practically creating training data from generative models touches various techniques from the field of computer graphics. In the short course, we present various useful techniques for the purpose and pitfalls in using them. As a practical relevant use-case, we demonstrate how to use a surrogate model of electron-matter interaction to create synthetic images of electron backscatter and secondary electron contrast.