Accelerating Deep Neural Networks for Real-time Data Selection for High-resolution Imaging Particle Detectors - INSPIRE

DataBETA

Accelerating Deep Neural Networks for Real-time Data Selection for High-resolution Imaging Particle Detectors

Yeon-Jae Jwa
(
- Columbia U.
)
,
Giuseppe Di Guglielmo
(
- Columbia U.
)
,
Luca P. Carloni
(
- Columbia U.
)
,
Georgia Karagiorgi
(
- Columbia U.
)

Jun, 2019

10 pages

Part of Proceedings, 2019 New York Scientific Data Summit: Data-Driven Discovery in Science and Industry (NYSDS) : New York, NY, USA, June 12-14, 2019

Contribution to:

- NYSDS

Published: Jun, 2019

e-Print:

2201.04740 [physics.ins-det]

DOI:

10.1109/NYSDS.2019.8909784

View in:

ADS Abstract Service

reference search6 citations

Citations per year

Abstract: (IEEE)

This paper presents the custom implementation, optimization, and performance evaluation of convolutional neural networks on field programmable gate arrays, for the purposes of accelerating deep neural network inference on large, two-dimensional image inputs. The targeted application is that of data selection for high-resolution particle imaging detectors, and in particular liquid argon time projection chamber detectors, such as that employed by the future Deep Underground Neutrino Experiment. We motivate this particular application based on the excellent performance of deep neural networks on classifying simulated raw data from the DUNE LArTPC, combined with the need for power-efficient data processing in the case of remote, long-term, and limited-access operating detector conditions.

Note:

10 pages, 5 figures, 8 tables

Detectors
Neutrino sources
Physics
Wires
Field programmable gate arrays
Neural networks
Real-time systems
field programmable gate arrays
image resolution
image sensors

References(34)

Figures(11)

[1]

Design and Construction of the MicroBooNE Detector

MicroBooNE

Collaboration

•

R. Acciarri
(
- Fermilab
)

et al.

- JINST 12 (2017) 02, P02017
•
e-Print:
- 1612.05824
•
DOI:
- 10.1088/1748-0221/12/02/P02017

[2]

The Single-Phase ProtoDUNE Technical Design Report

DUNE

Collaboration

•

B. Abi
(
- Oxford U.
)

et al.

e-Print:
- 1706.07081

[3]

The DUNE Far Detector Interim Design Report, Volume 2: Single-Phase Module

DUNE

Collaboration

•

B. Abi
(
- Oxford U.
)

et al.

e-Print:
- 1807.10327

[4]

Deep Underground Neutrino Experiment (DUNE) technical design report

[5]

Machine learning at the energy and intensity frontiers of particle physics

A. Radovic

[6]

Deep neural network for pixel-level electromagnetic particle identification in the MicroBooNE liquid argon time projection chamber

MicroBooNE

Collaboration

•

C. Adams
(
- Yale U.
)

et al.

- Phys.Rev.D 99 (2019) 9, 092001
•
e-Print:
- 1808.07269
•
DOI:
- 10.1103/PhysRevD.99.092001

[7]

Convolutional neural networks applied to neutrino events in a liquid argon time projection chamber

R. Acciarri

[8]

Scalable deep convolutional neural networks for sparse, locally dense liquid argon time projection chamber data

DeepLearnPhysics

Collaboration

•

Laura Dominé
(
- Stanford U. and
- SLAC
)

et al.

- Phys.Rev.D 102 (2020) 1, 012005
•
e-Print:
- 1903.05663
•
DOI:
- 10.1103/PhysRevD.102.012005

[9]

Efficient processing of deep neural networks: A tutorial and survey

V. Sze

[10]

FPGAs versus GPUs in data centers

B. Falsafi

[11]

Caffe: Convolutional architecture for fast feature embedding

Y. Jia

[12]

Tensorflow: A system for large-scale machine learning

M. Abadi

[13]

FPGA-accelerated machine learning inference as a service for particle physics computing

Javier Duarte
(
- Fermilab
)
,
Philip Harris
(
- MIT
)
,
Scott Hauck
(
- Washington U., Seattle
)
,
Burt Holzman
(
- Fermilab
)
,
Shih-Chieh Hsu
(
- Washington U., Seattle
)

et al.

- Comput.Softw.Big Sci. 3 (2019) 1, 13
•
e-Print:
- 1904.08986
•
DOI:
- 10.1007/s41781-019-0027-2

[14]

An introduction to high-level synthesis

P. Coussy

[15]

The Historical Supernovae

D.H. Clark
,
F.R. Stephenson

[15]

Proceedings of the Advanced Study Institute. Cambridge, England:

D. Dordrecht

[16]

LArSoft release redmine 06.60.00

- https://cdcvs.fnal.gov/redmine/projects/larsoft/wiki/ReleaseNotes066000

[17]

LArSoft release redmine 07.13.00

- https://cdcvs.fnal.gov/redmine/projects/larsoft/wiki/ReleaseNotes071300

[18]

dunetpc release redmine 07.13.00

- https://cdcvs.fnal.gov/redmine/projects/dunetpc

[19]

The Continuous Readout Stream of the MicroBooNE Liquid Argon Time Projection Chamber for Detection of Supernova Neutrinos

[20]

Very Deep Convolutional Networks for Large-Scale Image Recognition

e-Print:
- 1409.1556

[21]

ML-based triggering for DUNE

G. Ge
,
Y. Jwa
,
G. Karagiorgi

[22]

ImageNet classification with deep convolutional neural networks

- Neural Inform.Proc.Syst. 25 2012

[23]

The perceptron: a probabilistic model for information storage and organization in the brain

[24]

Deep Residual Learning for Image Recognition

e-Print:
- 1512.03385
•
DOI:
- 10.1109/CVPR.2016.90

1-25 of 34
1
2
25 / page