Würzburg 2018 – wissenschaftliches Programm

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

T: Fachverband Teilchenphysik

T 16: GRID Computing / Experimentelle Methoden I

T 16.2: Vortrag

Montag, 19. März 2018, 16:15–16:30, Z6 - SR 2.006

VISPA: Multi-user access to deep learning infrastructure for physics research — •Ralf Florian von Cube, Martin Erdmann, Benjamin Fischer, Robert Fischer, Erik Geiser, Christian Glaser, Thorben Quast, Marcel Rieger, Felix Schlüter, and Martin Urban — III. Physikalisches Institut A, RWTH Aachen University

The VISPA (Visual Physics Analysis) platform enables access to remote resources for performing physics analyses through a modern web-browser. A set of common tools to meet the demands of most physics data analysts is provided such that local software installations by users are not needed.

Recently, the VISPA cluster was extended by GeForce GTX 1080 cards and software to develop deep neural networks was installed. This enables users to explore modern methods of machine learning in their data analyses.

Access to resources is managed by a permission system, and allows experiment specific permission on dedicated resources.

The setup was successfully used in university Bachelor and Master courses and workshops with 100 participants.