Dresden 2026 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

TUT: Tutorien

TUT 3: Tutorial: Machine Learning Use Cases in Materials Science (joint session AKPIK/TUT)

TUT 3.2: Tutorial

Sunday, March 8, 2026, 16:05–16:45, HSZ/0003

A practical machine learning case study in materials science: Stumbling blocks, lucky breaks, and helpful colleagues — •Max Grossmann, •Malte Grunert, and Erich Runge — Institute of Physics and Institute of Micro- and Nanotechnologies, Technische Universität Ilmenau, 98693 Ilmenau, German

Machine learning projects in materials science are often exciting at first, but quickly encounter practical challenges. In this tutorial, we present a hands-on case study from our group and walk through the entire process, from the initial idea to building a dataset to training a working model. Rather than focusing solely on technical details, we highlight the stumbling blocks, unexpected insights, helpful colleagues, and lucky coincidences that shaped the project. We discuss the most important aspects of starting a machine learning project in physics or materials science, such as choosing a meaningful target property, designing a reliable dataset, avoiding common pitfalls, and identifying situations where simple approaches are as effective as advanced ones. Our goal is to provide an honest, accessible, and experience-driven introduction and guidance to researchers considering venturing into machine learning for the first time – the kind of guidance we wish we had when we started.

Keywords: Machine Learning; Materials Science; Case Study; Pitfalls; Best Practices