Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

HK: Fachverband Physik der Hadronen und Kerne

HK 44: Instrumentation IX

HK 44.5: Vortrag

Mittwoch, 20. März 2019, 17:45–18:00, HS 11

Using neural networks for event reconstuction at NeuLAND — •E. Hoemann, J. Mayer, P. Scholz, and A. Zilges — University of Cologne, Institute for Nuclear Physics

In various fields of modern data processing, neural networks play a key role. Popular applications like speech and face recognition are already part of our everyday lives. Where can we apply them in science for similar questions?

For the New Large Area Neutron Detector NeuLAND[1] we come across two classification problems: How many neutrons have interacted in the detector and which clusters are created through primary interactions? Humans need to reduce the data set to special quantities to gain information, but some underlying correlations could be missed in this process. In contrast, a neural network can use the whole data set, complex structures, and optimization algorithms to profit from these correlations.

The talk will address different approaches to construct a neural network for the mentioned classification problems and its performance in comparison to the conventional methods.

Supported by the BMBF(05P19PKFNA).

[1]Technical Report for the Design, Construction and Commissioning of NeuLAND, available at https://edms.cern.ch/document/1865739/1

100% | Bildschirmansicht | English Version | Kontakt/Impressum/Datenschutz
DPG-Physik > DPG-Verhandlungen > 2019 > München