Graph Neural Networks for low-energy event classification & reconstruction in IceCube
IceCube, a cubic-kilometer array of optical sensors built to detect atmospheric and astrophysical neutrinos between 1 GeV and 1 PeV, is deployed 1.45 km to 2.45 km below the surface of the ice sheet at the South Pole. The classification and reconstruction of events from the in-ice detectors play a c...
Main Authors: | , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , |
---|---|
Format: | Article in Journal/Newspaper |
Language: | English |
Published: |
Humboldt-Universität zu Berlin
2022
|
Subjects: | |
Online Access: | http://edoc.hu-berlin.de/18452/28339 https://nbn-resolving.org/urn:nbn:de:kobv:11-110-18452/28339-3 https://doi.org/10.1088/1748-0221/17/11/P11003 https://doi.org/10.18452/27684 |
id |
fthuberlin:oai:edoc.hu-berlin.de:18452/28339 |
---|---|
record_format |
openpolar |
institution |
Open Polar |
collection |
Open-Access-Publikationsserver der Humboldt-Universität: edoc-Server |
op_collection_id |
fthuberlin |
language |
English |
topic |
Analysis and statistical methods Data analysis Neutrino detectors Particle identification methods 520 Astronomie und zugeordnete Wissenschaften 530 Physik ddc:520 ddc:530 |
spellingShingle |
Analysis and statistical methods Data analysis Neutrino detectors Particle identification methods 520 Astronomie und zugeordnete Wissenschaften 530 Physik ddc:520 ddc:530 Abbasi, R. Ackermann, M. Adams, J. Aggarwal, N. Aguilar, J.A. Ahlers, M. Ahrens, M. Alameddine, J.M. Alves , A.A. Amin, N.M. Andeen, K. Anderson, T. Anton, G. Argüelles, C. Ashida, Y. Athanasiadou, S. Axani, S. Bai, X. Balagopal V., A. Baricevic, M. Barwick, S.W. Basu, V. Bay, R. Beatty, J.J. Becker, K.-H. Becker Tjus, J. Beise, J. Bellenghi, C. Benda, S. BenZvi, S. Berley, D. Bernardini, E. Besson, D.Z. Binder, G. Bindig, D. Blaufuss, E. Blot, S. Bontempo, F. Book, J.Y. Borowka, J. Boscolo Meneguolo, C. Böser, S. Botner, O. Böttcher, J. Bourbeau, E. Braun, J. Brinson, B. Brostean-Kaiser, J. Burley, R.T. Busse, R.S. Campana, M.A. Carnie-Bronca, E.G. Chen, C. Chen, Z. Chirkin, D. Choi, K. Clark, B.A. Classen, L. Coleman, A. Collin, G.H. Connolly, A. Conrad, J.M. Coppin, P. Correa, P. Countryman, S. Cowen, D.F. Cross, R. Dappen, C. Dave, P. De Clercq, C. DeLaunay, J.J. Delgado López, D. Dembinski, H. Deoskar, K. Desai, A. Desiati, P. de Vries, K.D. de Wasseige, G. DeYoung, T. Diaz, A. Díaz-Vélez, J.C. Dittmer, M. Dujmovic, H. DuVernois, M.A. Ehrhardt, T. Eller, P. Engel, R. Erpenbeck, H. Evans, J. Evenson, P.A. Fan, K.L. Fazely, A.R. Fedynitch, A. Feigl, N. Fiedlschuster, S. Fienberg, A.T. Finley, C. Fischer, L. Fox, D. Franckowiak, A. Kolanoski, Hermann Kowalski, Marek Graph Neural Networks for low-energy event classification & reconstruction in IceCube |
topic_facet |
Analysis and statistical methods Data analysis Neutrino detectors Particle identification methods 520 Astronomie und zugeordnete Wissenschaften 530 Physik ddc:520 ddc:530 |
description |
IceCube, a cubic-kilometer array of optical sensors built to detect atmospheric and astrophysical neutrinos between 1 GeV and 1 PeV, is deployed 1.45 km to 2.45 km below the surface of the ice sheet at the South Pole. The classification and reconstruction of events from the in-ice detectors play a central role in the analysis of data from IceCube. Reconstructing and classifying events is a challenge due to the irregular detector geometry, inhomogeneous scattering and absorption of light in the ice and, below 100 GeV, the relatively low number of signal photons produced per event. To address this challenge, it is possible to represent IceCube events as point cloud graphs and use a Graph Neural Network (GNN) as the classification and reconstruction method. The GNN is capable of distinguishing neutrino events from cosmic-ray backgrounds, classifying different neutrino event types, and reconstructing the deposited energy, direction and interaction vertex. Based on simulation, we provide a comparison in the 1 GeV–100 GeV energy range to the current state-of-the-art maximum likelihood techniques used in current IceCube analyses, including the effects of known systematic uncertainties. For neutrino event classification, the GNN increases the signal efficiency by 18% at a fixed background rate, compared to current IceCube methods. Alternatively, the GNN offers a reduction of the background (i.e. false positive) rate by over a factor 8 (to below half a percent) at a fixed signal efficiency. For the reconstruction of energy, direction, and interaction vertex, the resolution improves by an average of 13%–20% compared to current maximum likelihood techniques in the energy range of 1 GeV–30 GeV. The GNN, when run on a GPU, is capable of processing IceCube events at a rate nearly double of the median IceCube trigger rate of 2.7 kHz, which opens the possibility of using low energy neutrinos in online searches for transient events. Peer Reviewed |
format |
Article in Journal/Newspaper |
author |
Abbasi, R. Ackermann, M. Adams, J. Aggarwal, N. Aguilar, J.A. Ahlers, M. Ahrens, M. Alameddine, J.M. Alves , A.A. Amin, N.M. Andeen, K. Anderson, T. Anton, G. Argüelles, C. Ashida, Y. Athanasiadou, S. Axani, S. Bai, X. Balagopal V., A. Baricevic, M. Barwick, S.W. Basu, V. Bay, R. Beatty, J.J. Becker, K.-H. Becker Tjus, J. Beise, J. Bellenghi, C. Benda, S. BenZvi, S. Berley, D. Bernardini, E. Besson, D.Z. Binder, G. Bindig, D. Blaufuss, E. Blot, S. Bontempo, F. Book, J.Y. Borowka, J. Boscolo Meneguolo, C. Böser, S. Botner, O. Böttcher, J. Bourbeau, E. Braun, J. Brinson, B. Brostean-Kaiser, J. Burley, R.T. Busse, R.S. Campana, M.A. Carnie-Bronca, E.G. Chen, C. Chen, Z. Chirkin, D. Choi, K. Clark, B.A. Classen, L. Coleman, A. Collin, G.H. Connolly, A. Conrad, J.M. Coppin, P. Correa, P. Countryman, S. Cowen, D.F. Cross, R. Dappen, C. Dave, P. De Clercq, C. DeLaunay, J.J. Delgado López, D. Dembinski, H. Deoskar, K. Desai, A. Desiati, P. de Vries, K.D. de Wasseige, G. DeYoung, T. Diaz, A. Díaz-Vélez, J.C. Dittmer, M. Dujmovic, H. DuVernois, M.A. Ehrhardt, T. Eller, P. Engel, R. Erpenbeck, H. Evans, J. Evenson, P.A. Fan, K.L. Fazely, A.R. Fedynitch, A. Feigl, N. Fiedlschuster, S. Fienberg, A.T. Finley, C. Fischer, L. Fox, D. Franckowiak, A. Kolanoski, Hermann Kowalski, Marek |
author_facet |
Abbasi, R. Ackermann, M. Adams, J. Aggarwal, N. Aguilar, J.A. Ahlers, M. Ahrens, M. Alameddine, J.M. Alves , A.A. Amin, N.M. Andeen, K. Anderson, T. Anton, G. Argüelles, C. Ashida, Y. Athanasiadou, S. Axani, S. Bai, X. Balagopal V., A. Baricevic, M. Barwick, S.W. Basu, V. Bay, R. Beatty, J.J. Becker, K.-H. Becker Tjus, J. Beise, J. Bellenghi, C. Benda, S. BenZvi, S. Berley, D. Bernardini, E. Besson, D.Z. Binder, G. Bindig, D. Blaufuss, E. Blot, S. Bontempo, F. Book, J.Y. Borowka, J. Boscolo Meneguolo, C. Böser, S. Botner, O. Böttcher, J. Bourbeau, E. Braun, J. Brinson, B. Brostean-Kaiser, J. Burley, R.T. Busse, R.S. Campana, M.A. Carnie-Bronca, E.G. Chen, C. Chen, Z. Chirkin, D. Choi, K. Clark, B.A. Classen, L. Coleman, A. Collin, G.H. Connolly, A. Conrad, J.M. Coppin, P. Correa, P. Countryman, S. Cowen, D.F. Cross, R. Dappen, C. Dave, P. De Clercq, C. DeLaunay, J.J. Delgado López, D. Dembinski, H. Deoskar, K. Desai, A. Desiati, P. de Vries, K.D. de Wasseige, G. DeYoung, T. Diaz, A. Díaz-Vélez, J.C. Dittmer, M. Dujmovic, H. DuVernois, M.A. Ehrhardt, T. Eller, P. Engel, R. Erpenbeck, H. Evans, J. Evenson, P.A. Fan, K.L. Fazely, A.R. Fedynitch, A. Feigl, N. Fiedlschuster, S. Fienberg, A.T. Finley, C. Fischer, L. Fox, D. Franckowiak, A. Kolanoski, Hermann Kowalski, Marek |
author_sort |
Abbasi, R. |
title |
Graph Neural Networks for low-energy event classification & reconstruction in IceCube |
title_short |
Graph Neural Networks for low-energy event classification & reconstruction in IceCube |
title_full |
Graph Neural Networks for low-energy event classification & reconstruction in IceCube |
title_fullStr |
Graph Neural Networks for low-energy event classification & reconstruction in IceCube |
title_full_unstemmed |
Graph Neural Networks for low-energy event classification & reconstruction in IceCube |
title_sort |
graph neural networks for low-energy event classification & reconstruction in icecube |
publisher |
Humboldt-Universität zu Berlin |
publishDate |
2022 |
url |
http://edoc.hu-berlin.de/18452/28339 https://nbn-resolving.org/urn:nbn:de:kobv:11-110-18452/28339-3 https://doi.org/10.1088/1748-0221/17/11/P11003 https://doi.org/10.18452/27684 |
geographic |
South Pole |
geographic_facet |
South Pole |
genre |
Ice Sheet South pole |
genre_facet |
Ice Sheet South pole |
op_relation |
http://edoc.hu-berlin.de/18452/28339 urn:nbn:de:kobv:11-110-18452/28339-3 doi:10.1088/1748-0221/17/11/P11003 http://dx.doi.org/10.18452/27684 1748-0221 |
op_rights |
(CC BY 4.0) Attribution 4.0 International https://creativecommons.org/licenses/by/4.0/ |
op_doi |
https://doi.org/10.1088/1748-0221/17/11/P1100310.18452/27684 |
_version_ |
1784894407655292928 |
spelling |
fthuberlin:oai:edoc.hu-berlin.de:18452/28339 2023-12-10T09:49:45+01:00 Graph Neural Networks for low-energy event classification & reconstruction in IceCube Abbasi, R. Ackermann, M. Adams, J. Aggarwal, N. Aguilar, J.A. Ahlers, M. Ahrens, M. Alameddine, J.M. Alves , A.A. Amin, N.M. Andeen, K. Anderson, T. Anton, G. Argüelles, C. Ashida, Y. Athanasiadou, S. Axani, S. Bai, X. Balagopal V., A. Baricevic, M. Barwick, S.W. Basu, V. Bay, R. Beatty, J.J. Becker, K.-H. Becker Tjus, J. Beise, J. Bellenghi, C. Benda, S. BenZvi, S. Berley, D. Bernardini, E. Besson, D.Z. Binder, G. Bindig, D. Blaufuss, E. Blot, S. Bontempo, F. Book, J.Y. Borowka, J. Boscolo Meneguolo, C. Böser, S. Botner, O. Böttcher, J. Bourbeau, E. Braun, J. Brinson, B. Brostean-Kaiser, J. Burley, R.T. Busse, R.S. Campana, M.A. Carnie-Bronca, E.G. Chen, C. Chen, Z. Chirkin, D. Choi, K. Clark, B.A. Classen, L. Coleman, A. Collin, G.H. Connolly, A. Conrad, J.M. Coppin, P. Correa, P. Countryman, S. Cowen, D.F. Cross, R. Dappen, C. Dave, P. De Clercq, C. DeLaunay, J.J. Delgado López, D. Dembinski, H. Deoskar, K. Desai, A. Desiati, P. de Vries, K.D. de Wasseige, G. DeYoung, T. Diaz, A. Díaz-Vélez, J.C. Dittmer, M. Dujmovic, H. DuVernois, M.A. Ehrhardt, T. Eller, P. Engel, R. Erpenbeck, H. Evans, J. Evenson, P.A. Fan, K.L. Fazely, A.R. Fedynitch, A. Feigl, N. Fiedlschuster, S. Fienberg, A.T. Finley, C. Fischer, L. Fox, D. Franckowiak, A. Kolanoski, Hermann Kowalski, Marek 2022-11-04 application/pdf http://edoc.hu-berlin.de/18452/28339 https://nbn-resolving.org/urn:nbn:de:kobv:11-110-18452/28339-3 https://doi.org/10.1088/1748-0221/17/11/P11003 https://doi.org/10.18452/27684 eng eng Humboldt-Universität zu Berlin http://edoc.hu-berlin.de/18452/28339 urn:nbn:de:kobv:11-110-18452/28339-3 doi:10.1088/1748-0221/17/11/P11003 http://dx.doi.org/10.18452/27684 1748-0221 (CC BY 4.0) Attribution 4.0 International https://creativecommons.org/licenses/by/4.0/ Analysis and statistical methods Data analysis Neutrino detectors Particle identification methods 520 Astronomie und zugeordnete Wissenschaften 530 Physik ddc:520 ddc:530 article doc-type:article publishedVersion 2022 fthuberlin https://doi.org/10.1088/1748-0221/17/11/P1100310.18452/27684 2023-11-12T23:35:42Z IceCube, a cubic-kilometer array of optical sensors built to detect atmospheric and astrophysical neutrinos between 1 GeV and 1 PeV, is deployed 1.45 km to 2.45 km below the surface of the ice sheet at the South Pole. The classification and reconstruction of events from the in-ice detectors play a central role in the analysis of data from IceCube. Reconstructing and classifying events is a challenge due to the irregular detector geometry, inhomogeneous scattering and absorption of light in the ice and, below 100 GeV, the relatively low number of signal photons produced per event. To address this challenge, it is possible to represent IceCube events as point cloud graphs and use a Graph Neural Network (GNN) as the classification and reconstruction method. The GNN is capable of distinguishing neutrino events from cosmic-ray backgrounds, classifying different neutrino event types, and reconstructing the deposited energy, direction and interaction vertex. Based on simulation, we provide a comparison in the 1 GeV–100 GeV energy range to the current state-of-the-art maximum likelihood techniques used in current IceCube analyses, including the effects of known systematic uncertainties. For neutrino event classification, the GNN increases the signal efficiency by 18% at a fixed background rate, compared to current IceCube methods. Alternatively, the GNN offers a reduction of the background (i.e. false positive) rate by over a factor 8 (to below half a percent) at a fixed signal efficiency. For the reconstruction of energy, direction, and interaction vertex, the resolution improves by an average of 13%–20% compared to current maximum likelihood techniques in the energy range of 1 GeV–30 GeV. The GNN, when run on a GPU, is capable of processing IceCube events at a rate nearly double of the median IceCube trigger rate of 2.7 kHz, which opens the possibility of using low energy neutrinos in online searches for transient events. Peer Reviewed Article in Journal/Newspaper Ice Sheet South pole Open-Access-Publikationsserver der Humboldt-Universität: edoc-Server South Pole |