Manchester Particle Physics Seminars, Colloquia and Meetings

Martin Erdmann - Deep Learning in Particle Physics Research

Friday, 3 November 2017 from to (Europe/London)
at Schuster ( 54-6-6.40 - Niels Bohr Common Room )
The success of deep learning in handwriting, speech and image
recognition is currently leading to rethinking in basic research. The
direct transfer of deep neural networks, e.g. picture-based
reconstruction of events by convolutional network architectures, works
impressively. It became evident that deep networks extract information
from the data and use it for predictions that were not immediately
accessible before. Therefore, the interested community currently focuses
on modelling particle physics into deep network architectures. This talk
provides an introduction to deep learning methods, presents different
network architectures and their applications in particle physics, and
discusses ways to introspect network precision and causality.