ML-SANS-Classification
| Topic | 23 |
| Main supervisor | S.Förster (s.foerster@fz-juelich.de) |
| MLZ institution | FZJ |
| Local supervisor 1 | tbd |
| Institution | XENOCS |
| Local supervisor 2 | – |
| Institution | – |
| Local supervisor 3 | – |
| Institution | – |
| Local supervisor 4 | – |
| Institution | – |
| Title | Towards Polarization Analysis for Quasi Elastic Neutron Scattering (PA-QENS) |
| Description | High-throughput SANS instruments such as KWS-1 and KWS-2 produce 2D-SANS images at high rate. The SANS-patterns may be complex such that a detailed data analysis will only be possible after the experiments.
The project will develop a machine-learning based software that recognizes and classifies a large number of classes of the most common structures with their isotropic and anisotropic SANS-patterns. After classification, the software will further allow to model the data with 2D- or 1D-fits using GPUs for rapid calculations. The project will be conducted with XENOCS, a company that sells X-ray equipment where a large number of small-angle data have to be analyzed. Students will become familiar with the software development and machine learning tools, and develop a modern portfolio of skills in both academic and industrial settings. |