Regensburg 2025 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
CPP: Fachverband Chemische Physik und Polymerphysik
CPP 30: Emerging Topics in Chemical and Polymer Physics, New Instruments and Methods I
CPP 30.1: Hauptvortrag
Mittwoch, 19. März 2025, 16:15–16:45, H34
Challenges and Opportunities in Bringing Machine Learning to a Synchrotron — •Alexander Hexemer1, Tanny Chavez1, Wiebke Köpp1, Dylan McReynolds1, Stephan Roth2, Tim Snow3, and Sharif Ahmed3 — 1Lawrence Berkeley National Lab, Berkeley, CA 94720 — 2DESY, Hamburg, Germany — 3Diamond Light Source, Didcot, UK
Artificial intelligence (AI) and machine learning (ML) are transforming scientific research, offering innovative solutions to longstanding data collection, analysis, and interpretation challenges. Synchrotron facilities, which generate vast amounts of complex, high-dimensional data, present unique opportunities to leverage ML to advance materials science. Building on this potential, significant progress is being made at the Advanced Light Source (ALS) to integrate ML tools into various synchrotron applications, including tomography segmentation, autonomous scattering analysis, and multimodal data fusion. Efforts are focused on implementing ML as a service, simplifying adoption by providing web-based solutions designed for seamless use across facilities. These tools aim to enable reliable and scalable ML applications for tasks such as the segmentation of complex 3D tomography datasets and automated experimental feedback in scattering experiments. This talk will explore the evolving role of ML at synchrotrons while addressing key challenges.
Keywords: Scattering; Scientific User facility; Machine Learning; Generative AI; Tomography