Dresden 2026 – scientific programme

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

MM: Fachverband Metall- und Materialphysik

MM 5: Topical Session: Physics-driven Artificial Intelligence for Materials I

MM 5.1: Topical Talk

Monday, March 9, 2026, 10:15–10:45, SCH/A251

Machine Learning for Materials Discovery: from Big Data to Predictive Insights — •Silvana Botti — Research Center Future Energy Materials and Systems and Interdisciplinary Centre for Advanced Materials Simulation, Ruhr University Bochum, Universitätsstraße 150, D-44801 Bochum, Germany

Machine learning (ML) models for materials science are rapidly evolving, driven by large-scale, high-quality datasets and innovative neural network architectures. This talk explores critical challenges in improving the accuracy and reliability of complex ML models, examining the interplay between the quality and quantity of training data and model performance across material properties. Recent advances have been marked by the creation of extensive FAIR databases, such as Alexandria (https://alexandria.icams.rub.de/), which provides over 7 million density-functional theory calculations spanning periodic compounds of various dimensionalities. These comprehensive datasets enable systematic investigation of the relationship between training data volume/quality and model accuracy.

J. Schmidt, T.F.T. Cerqueira, A.H. Romero, A. Loew, F. Jäger, H.-C. Wang, S. Botti, M.A.L. Marques, Improving machine-learning models in materials science through large datasets, Mater. Today Phys. 48, 101560 (2024).

Keywords: materials databases; machine learning; density functional theory; universal machine learning interatomic potentials