Deep Neural Networks for Physics Analysis on low-level whole-detector data at the LHC

Article
Report number	arXiv:1711.03573
Title	Deep Neural Networks for Physics Analysis on low-level whole-detector data at the LHC
Author(s)	Bhimji, Wahid (LBNL, Berkeley) ; Farrell, Steven Andrew (LBNL, Berkeley) ; Kurth, Thorsten (LBNL, Berkeley) ; Paganini, Michela (LBNL, Berkeley ; Yale U.) ; Prabhat (LBNL, Berkeley) ; Racah, Evan (LBNL, Berkeley)
Publication	2018-10-18
Imprint	2017-11-09
Number of pages	6
Note	Presented at ACAT 2017 Conference, Submitted to J. Phys. Conf. Ser
In:	J. Phys.: Conf. Ser. 1085 (2018) 042034
In:	18th International Workshop on Advanced Computing and Analysis Techniques in Physics Research, Seattle, WA, USA, 21 - 25 Aug 2017, pp.042034
DOI	10.1088/1742-6596/1085/4/042034
Subject category	physics.data-an ; Other Fields of Physics ; cs.LG ; Computing and Computers ; cs.DC ; Computing and Computers ; hep-ex ; Particle Physics - Experiment
Accelerator/Facility, Experiment	CERN LHC ; ATLAS
Abstract	There has been considerable recent activity applying deep convolutional neural nets (CNNs) to data from particle physics experiments. Current approaches on ATLAS/CMS have largely focussed on a subset of the calorimeter, and for identifying objects or particular particle types. We explore approaches that use the entire calorimeter, combined with track information, for directly conducting physics analyses: i.e. classifying events as known-physics background or new-physics signals. We use an existing RPV-Supersymmetry analysis as a case study and explore CNNs on multi-channel, high-resolution sparse images: applied on GPU and multi-node CPU architectures (including Knights Landing (KNL) Xeon Phi nodes) on the Cori supercomputer at NERSC.
Copyright/License	arXiv nonexclusive-distrib. 1.0 publication: (License: CC-BY-3.0)