Machine-learning methods are bridging the gap between macroscopically relevant and quantum-mechanical length scales. Machine learning has the potential to transform materials science by expediting discovery, enhancing prediction capabilities, optimizing processes, enabling data-driven design, and automating analysis tasks. At the same time, known physical laws and constraints can enhance machine learning models by guiding feature engineering, integrating prior knowledge, augmenting data, promoting interpretability, and facilitating model calibration. This SIG will bring together computer scientists and theorists and experimentalist from the sciences to develop powerful and robust models in materials science and identify future directions of research.

Members

• Georg Madsen
• Ulrike Diebold
• Thomas Gärtner
• Florian Libisch
• Esther Heid
• Andreas Grüneis