About this Abstract |
| Meeting |
MS&T25: Materials Science & Technology
|
| Symposium
|
Enhancing the Accessibility of Machine Learning-Enabled Experiments
|
| Presentation Title |
Accelerating Scientific Discovery with Machine Learning: Data Analysis for Computational Beamlines |
| Author(s) |
Tanny Chavez, Xiaoya Chong, Bowen Zheng, Anas Nassar, Monika Choudhary, Wiebke Koepp, Dylan McReynolds, Slavomir Nemsak, Alexander Hexemer |
| On-Site Speaker (Planned) |
Tanny Chavez |
| Abstract Scope |
As user facilities push the boundaries of experimental science, the need for intelligent, real-time computational tools becomes increasingly critical. This presentation introduces a suite of machine learning (ML) frameworks tailored to the unique challenges of data-intensive beamline science. These include automated data labeling through clustering and similarity search, interactive latent space navigation for real-time feedback during experiments, and cross-facility model adaptation to promote reusability and broader applicability. Deployed at the Advanced Light Source (ALS) and collaborating facilities, these tools support in situ analysis of complex, intricate scientific datasets, supporting researchers to make informed experimental decisions on-the-fly. We also highlight efforts to integrate physics-aware ML for the interpretation of multimodal data, such as X-ray scattering and X-ray Photoelectron Spectroscopy. Together, these solutions form the foundation of a scalable, reproducible, and facility-agnostic Machine Learning as a Service (MLaaS) ecosystem that closes the loop between data acquisition and insight. |