skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Pulling out all the tops with computer vision and deep learning

Abstract

We apply computer vision with deep learning — in the form of a convolutional neural network (CNN) — to build a highly effective boosted top tagger. Previous work (the “DeepTop” tagger of Kasieczka et al) has shown that a CNN-based top tagger can achieve comparable performance to state-of-the-art conventional top taggers based on high-level inputs. Here, we introduce a number of improvements to the DeepTop tagger, including architecture, training, image preprocessing, sample size and color pixels. Our final CNN top tagger outperforms BDTs based on high-level inputs by a factor of ~ 2–3 or more in background rejection, over a wide range of tagging efficiencies and fiducial jet selections. As reference points, we achieve a QCD background rejection factor of 500 (60) at 50%top tagging efficiency for fully-merged (non-merged) top jets with pT in the 800–900 GeV (350–450 GeV) range. Our CNN can also be straightforwardly extended to the classification of other types of jets, and the lessons learned here may be useful to others designing their own deep NNs for LHC applications.

Authors:
 [1];  [1]
  1. Rutgers Univ., Piscataway, NJ (United States)
Publication Date:
Research Org.:
Rutgers Univ., Piscataway, NJ (United States)
Sponsoring Org.:
USDOE
OSTI Identifier:
1483579
Grant/Contract Number:  
SC0010008
Resource Type:
Journal Article: Accepted Manuscript
Journal Name:
Journal of High Energy Physics (Online)
Additional Journal Information:
Journal Volume: 2018; Journal Issue: 10; Journal ID: ISSN 1029-8479
Publisher:
Springer Berlin
Country of Publication:
United States
Language:
English
Subject:
97 MATHEMATICS AND COMPUTING; Jets

Citation Formats

Macaluso, Sebastian, and Shih, David. Pulling out all the tops with computer vision and deep learning. United States: N. p., 2018. Web. doi:10.1007/JHEP10(2018)121.
Macaluso, Sebastian, & Shih, David. Pulling out all the tops with computer vision and deep learning. United States. doi:10.1007/JHEP10(2018)121.
Macaluso, Sebastian, and Shih, David. Thu . "Pulling out all the tops with computer vision and deep learning". United States. doi:10.1007/JHEP10(2018)121. https://www.osti.gov/servlets/purl/1483579.
@article{osti_1483579,
title = {Pulling out all the tops with computer vision and deep learning},
author = {Macaluso, Sebastian and Shih, David},
abstractNote = {We apply computer vision with deep learning — in the form of a convolutional neural network (CNN) — to build a highly effective boosted top tagger. Previous work (the “DeepTop” tagger of Kasieczka et al) has shown that a CNN-based top tagger can achieve comparable performance to state-of-the-art conventional top taggers based on high-level inputs. Here, we introduce a number of improvements to the DeepTop tagger, including architecture, training, image preprocessing, sample size and color pixels. Our final CNN top tagger outperforms BDTs based on high-level inputs by a factor of ~ 2–3 or more in background rejection, over a wide range of tagging efficiencies and fiducial jet selections. As reference points, we achieve a QCD background rejection factor of 500 (60) at 50%top tagging efficiency for fully-merged (non-merged) top jets with pT in the 800–900 GeV (350–450 GeV) range. Our CNN can also be straightforwardly extended to the classification of other types of jets, and the lessons learned here may be useful to others designing their own deep NNs for LHC applications.},
doi = {10.1007/JHEP10(2018)121},
journal = {Journal of High Energy Physics (Online)},
issn = {1029-8479},
number = 10,
volume = 2018,
place = {United States},
year = {2018},
month = {10}
}

Journal Article:
Free Publicly Available Full Text
Publisher's Version of Record

Save / Share: