Smita Chakraborty: AI-SAXS – Decoding Structural Complexity with Intelligent Scattering Analysis

The AI-SAXS project explores the possibilities of using machine learning to interpret the information in the small angle X-ray scattering curves together with project partners AstraZeneca and TetraPak. The model developed within the project will be trained on synthetic data and real experimental data from the Max IV synchrotron laboratory in Lund, Sweden. 

By training the model on thousands of SAXS spectra of representative models, it learns to interpret real samples that are of interest for the experiments in the project. One of the goals is to make the models and working methods as generalizable as possible given the diversity of the project focus samples which are primarily cellulose for TetraPak and lipid nanoparticles in the case of AstraZeneca. Both the work process and type of neural networks are evaluated in the project to find the best methods for these different materials.

Smita Chakraborty is a researcher at the Computer Science Department at RISE. She earned her Ph.D. in Theoretical Particle Physics from Lund University in 2022, where her doctoral thesis focused on extending the Lund String Model for heavy ion collisions at particle colliders such as CERN. Her current research interests are physics-informed machine learning, ML for earth observation, and quantum computation.