témavezető: Brockhauser Sándor
helyszín (magyar oldal): SZTE helyszín rövidítés: SZTE
A kutatási téma leírása:
European XFEL is an international non-profit company located in the Hamburg area in
Germany. It operates a 3.4 km-long X-ray laser which generates ultrashort X-ray flashes
—27 000 times per second and with a brilliance that is a billion times higher than that of
the best conventional X-ray radiation sources. It produces X-rays of unique quality for
studies in physics, chemistry, the life sciences, materials research and other disciplines.
The diverse scientific facilities at European XFEL enable scientists from across the
globe to carry out a wide range of experimental techniques. Early user operation has
started in September 2017.
Given the brilliance of third-generation synchrotron sources and the consequent
diversity of possible experimental set-ups it is perhaps unsurprising that the scientific
data produced come in a huge variety of formats and sizes. The primary role of these
facilities is to allow users to analyse samples on state-of-the-art experiments and leave
with data that can easily be processed and prepared for publication.
A critical part of carrying out photon science experiments is to have quick feedback on
the correctness of experimental conditions during the experiment, including the crucial
question of the sample being physically in the photon beam. This is currently done by
hit detection software (which helps diagnosing if the beam has hit the sample) that
generally requires a trained scientist to monitor the output to decide if the detector
image indicates a successful 'hit'. In this project, we explore how emerging machine
learning methodology can be used to provide an automatic assessment of the situation
to make conducting the experiment more effective, and how it can be used in later
stages of the data analysis chain.
Topics: machine learning (initially on 2d images), Data Analytics & Machine Learning
előírt nyelvtudás: English felvehető hallgatók száma: 1