Choose timezone
Your profile timezone:
ML-Reflectometry workshop
→
Europe/Berlin
UYG 02 22 - Pforte (MLZ campus Garching)
,
Description
The Data Evaluation Group and the Data Driven Discovery Group are pleased to announce a one day Machine Learning supported - neutron reflectometry workshop in Garching.
This hands-on workshop introduces conventional and machine-learning-supported approaches to neutron reflectometry data analysis. It is intended primarily for our MLZ User community and all MLZ researchers (instrument scientists, students etc), who are interested in reflectometry and scientific machine learning.
The morning session will introduce the basic principles of neutron reflectometry and demonstrate the analysis of experimental data using the conventional fitting software Anaklasis.
The afternoon session will provide a short recap of supervised machine learning, neural networks, and their application to parameter estimation from reflectivity data. The main part of the afternoon will introduce state-of-the-art probabilistic analysis methods based on advanced neural network models.
Using prepared notebooks and simulated and experimental reflectometry data, participants will investigate posterior parameter distributions, the role of prior information, parameter correlations, and ambiguities in the inverse problem.
No advanced machine-learning experience is required. The workshop will use prepared Python notebooks and guided exercises.
Neelima Paul