2017 Meeting of the APS Division of Particles and Fields (DPF 2017)

Exploring end-to-end image-based deep learning for particle & event classification

31 Jul 2017, 19:07

1m

Reception Area (Fermi National Accelerator Laboratory)

Poster Computing, Analysis Tools and Data Handling Poster Session and Reception

Speaker

Michael Andrews (Carnegie Mellon University)

Description

An essential part of new physics searches at the Large Hadron Collider at CERN involves event classification, or distinguishing signal decays from one of its many background look-alikes. Traditional techniques have relied on reconstructed particle candidates and their physical attributes. However, such reconstructed data are the result of a long, potentially lossy process of forcing the raw sensor data into progressively more physically intuitive quantities. Meanwhile, powerful image-based machine learning algorithms have emerged that are able to directly digest raw sensory data and output a prediction---so called end-to-end deep learning classifiers. We explore the use of such algorithms to perform physics classification using raw sensory data from the CMS detector. As proof of concept, we classify photon- versus electron-induced ECAL showers and compare the performance of using raw sensor data versus shower shape data.