4–5 Nov 2021
Helmholtz-Zentrum Dresden-Rossendorf
Europe/Berlin timezone

Reconstruction of SAXS Data using Neural Networks

4 Nov 2021, 16:05
20m
Building 106, Hörsaal room 255 (Helmholtz-Zentrum Dresden-Rossendorf)

Building 106, Hörsaal room 255

Helmholtz-Zentrum Dresden-Rossendorf

Bautzner Landstraße 400 01328 Dresden Zoom-Link: https://us02web.zoom.us/j/86275700441?pwd=Q29VUlF2MXVnQUlVL1M0TU9KZmd4QT09 Meeting ID: 862 7570 0441 Passcode: PuBTU9
Work in progress Talks

Speaker

Erik Thiessenhusen (HZDR)

Description

We aim to simplify the process of reconstructing electron densities from SAXS images by employing a special Neural Network architecture, the conditional Invertible Neural Network (cINN). The only requirement is a simulation from electron density to diffraction image to generate a training dataset. Once trained, it can make accurate and fast (ms range) predictions on simulated and experimental data and furthermore resolve ambiguities resulting from the phase problem. Some challenges remain though, since we cannot differentiate between accurate predictions and false predictions from experimental data not covered by the training dataset (out-of-distribution data) as the output does not convey the degree of certainty of the prediction made by the cINN.

Primary author

Presentation materials

There are no materials yet.