JUNIPR: a framework for unsupervised machine learning in particle physics
Abstract
In applications of machine learning to particle physics, a persistent challenge is how to go beyond discrimination to learn about the underlying physics. To this end, a powerful tool would be a framework for unsupervised learning, where the machine learns the intricate high-dimensional contours of the data upon which it is trained, without reference to pre-established labels. In order to approach such a complex task, an unsupervised network must be structured intelligently, based on a qualitative understanding of the data. In this paper, we scaffold the neural network’s architecture around a leading-order model of the physics underlying the data. In addition to making unsupervised learning tractable, this design actually alleviates existing tensions between performance and interpretability. We call the framework JUNIPR: “Jets from UNsupervised Interpretable PRobabilistic models”. In this approach, the set of particle momenta composing a jet are clustered into a binary tree that the neural network examines sequentially. Training is unsupervised and unrestricted: the network could decide that the data bears little correspondence to the chosen tree structure. However, when there is a correspondence, the network’s output along the tree has a direct physical interpretation. JUNIPR models can perform discrimination tasks, through the statistically optimal likelihood-ratio test, andmore »
- Authors:
- Publication Date:
- Research Org.:
- Harvard Univ., Cambridge, MA (United States)
- Sponsoring Org.:
- USDOE Office of Science (SC)
- OSTI Identifier:
- 1619354
- Alternate Identifier(s):
- OSTI ID: 1612013
- Grant/Contract Number:
- SC0013607
- Resource Type:
- Published Article
- Journal Name:
- European Physical Journal. C, Particles and Fields
- Additional Journal Information:
- Journal Name: European Physical Journal. C, Particles and Fields Journal Volume: 79 Journal Issue: 2; Journal ID: ISSN 1434-6044
- Publisher:
- Springer
- Country of Publication:
- Germany
- Language:
- English
- Subject:
- 72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS; Physics
Citation Formats
Andreassen, Anders, Feige, Ilya, Frye, Christopher, and Schwartz, Matthew D. JUNIPR: a framework for unsupervised machine learning in particle physics. Germany: N. p., 2019.
Web. doi:10.1140/epjc/s10052-019-6607-9.
Andreassen, Anders, Feige, Ilya, Frye, Christopher, & Schwartz, Matthew D. JUNIPR: a framework for unsupervised machine learning in particle physics. Germany. https://doi.org/10.1140/epjc/s10052-019-6607-9
Andreassen, Anders, Feige, Ilya, Frye, Christopher, and Schwartz, Matthew D. Fri .
"JUNIPR: a framework for unsupervised machine learning in particle physics". Germany. https://doi.org/10.1140/epjc/s10052-019-6607-9.
@article{osti_1619354,
title = {JUNIPR: a framework for unsupervised machine learning in particle physics},
author = {Andreassen, Anders and Feige, Ilya and Frye, Christopher and Schwartz, Matthew D.},
abstractNote = {In applications of machine learning to particle physics, a persistent challenge is how to go beyond discrimination to learn about the underlying physics. To this end, a powerful tool would be a framework for unsupervised learning, where the machine learns the intricate high-dimensional contours of the data upon which it is trained, without reference to pre-established labels. In order to approach such a complex task, an unsupervised network must be structured intelligently, based on a qualitative understanding of the data. In this paper, we scaffold the neural network’s architecture around a leading-order model of the physics underlying the data. In addition to making unsupervised learning tractable, this design actually alleviates existing tensions between performance and interpretability. We call the framework JUNIPR: “Jets from UNsupervised Interpretable PRobabilistic models”. In this approach, the set of particle momenta composing a jet are clustered into a binary tree that the neural network examines sequentially. Training is unsupervised and unrestricted: the network could decide that the data bears little correspondence to the chosen tree structure. However, when there is a correspondence, the network’s output along the tree has a direct physical interpretation. JUNIPR models can perform discrimination tasks, through the statistically optimal likelihood-ratio test, and they permit visualizations of discrimination power at each branching in a jet’s tree. Additionally, JUNIPR models provide a probability distribution from which events can be drawn, providing a data-driven Monte Carlo generator. As a third application, JUNIPR models can reweight events from one (e.g. simulated) data set to agree with distributions from another (e.g. experimental) data set.},
doi = {10.1140/epjc/s10052-019-6607-9},
journal = {European Physical Journal. C, Particles and Fields},
number = 2,
volume = 79,
place = {Germany},
year = {Fri Feb 01 00:00:00 EST 2019},
month = {Fri Feb 01 00:00:00 EST 2019}
}
https://doi.org/10.1140/epjc/s10052-019-6607-9
Web of Science
Works referenced in this record:
The anti- k t jet clustering algorithm
journal, April 2008
- Cacciari, Matteo; Salam, Gavin P.; Soyez, Gregory
- Journal of High Energy Physics, Vol. 2008, Issue 04
Learning representations by back-propagating errors
journal, October 1986
- Rumelhart, David E.; Hinton, Geoffrey E.; Williams, Ronald J.
- Nature, Vol. 323, Issue 6088
Singularities in the physical region
journal, July 1965
- Coleman, S.; Norton, R. E.
- Il Nuovo Cimento, Vol. 38, Issue 1
Soft gluons and factorization
journal, October 1988
- Collins, John C.; Soper, Davison E.; Sterman, George
- Nuclear Physics B, Vol. 308, Issue 4
Finding physics signals with event deconstruction
journal, May 2014
- Soper, Davison E.; Spannowsky, Michael
- Physical Review D, Vol. 89, Issue 9
How much information is in a jet?
journal, June 2017
- Datta, Kaustuv; Larkoski, Andrew
- Journal of High Energy Physics, Vol. 2017, Issue 6
Speech recognition with deep recurrent neural networks
conference, May 2013
- Graves, Alex; Mohamed, Abdel-rahman; Hinton, Geoffrey
- ICASSP 2013 - 2013 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
(Machine) learning to do more with less
journal, February 2018
- Cohen, Timothy; Freytsis, Marat; Ostdiek, Bryan
- Journal of High Energy Physics, Vol. 2018, Issue 2
CaloGAN: Simulating 3D high energy particle showers in multilayer electromagnetic calorimeters with generative adversarial networks
journal, January 2018
- Paganini, Michela; de Oliveira, Luke; Nachman, Benjamin
- Physical Review D, Vol. 97, Issue 1
Finding physics signals with shower deconstruction
journal, October 2011
- Soper, Davison E.; Spannowsky, Michael
- Physical Review D, Vol. 84, Issue 7
Classification without labels: learning from mixed samples in high energy physics
journal, October 2017
- Metodiev, Eric M.; Nachman, Benjamin; Thaler, Jesse
- Journal of High Energy Physics, Vol. 2017, Issue 10
Longitudinally-invariant k⊥-clustering algorithms for hadron-hadron collisions
journal, September 1993
- Catani, S.; Dokshitzer, Yu. L.; Seymour, M. H.
- Nuclear Physics B, Vol. 406, Issue 1-2
PYTHIA 6.4 physics and manual
journal, May 2006
- Sjöstrand, Torbjörn; Mrenna, Stephen; Skands, Peter
- Journal of High Energy Physics, Vol. 2006, Issue 05
Pileup Mitigation with Machine Learning (PUMML)
journal, December 2017
- Komiske, Patrick T.; Metodiev, Eric M.; Nachman, Benjamin
- Journal of High Energy Physics, Vol. 2017, Issue 12
Factorization for short distance hadron-hadron scattering
journal, January 1985
- Collins, John C.; Soper, Davison E.; Sterman, George
- Nuclear Physics B, Vol. 261
Seeing in Color: Jet Superstructure
journal, July 2010
- Gallicchio, Jason; Schwartz, Matthew D.
- Physical Review Letters, Vol. 105, Issue 2
Deep learning in color: towards automated quark/gluon jet discrimination
journal, January 2017
- Komiske, Patrick T.; Metodiev, Eric M.; Schwartz, Matthew D.
- Journal of High Energy Physics, Vol. 2017, Issue 1
Deep-learning top taggers or the end of QCD?
journal, May 2017
- Kasieczka, Gregor; Plehn, Tilman; Russell, Michael
- Journal of High Energy Physics, Vol. 2017, Issue 5
Multivariate discrimination and the Higgs+W/Z search
journal, April 2011
- Gallicchio, Jason; Huth, John; Kagan, Michael
- Journal of High Energy Physics, Vol. 2011, Issue 4
FastJet user manual: (for version 3.0.2)
journal, March 2012
- Cacciari, Matteo; Salam, Gavin P.; Soyez, Gregory
- The European Physical Journal C, Vol. 72, Issue 3
Jet-images: computer vision inspired techniques for jet tagging
journal, February 2015
- Cogan, Josh; Kagan, Michael; Strauss, Emanuel
- Journal of High Energy Physics, Vol. 2015, Issue 2
Fuzzy jets
journal, June 2016
- Mackey, Lester; Nachman, Benjamin; Schwartzman, Ariel
- Journal of High Energy Physics, Vol. 2016, Issue 6
Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation
conference, January 2014
- Cho, Kyunghyun; van Merrienboer, Bart; Gulcehre, Caglar
- Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP)
Novel jet observables from machine learning
journal, March 2018
- Datta, Kaustuv; Larkoski, Andrew J.
- Journal of High Energy Physics, Vol. 2018, Issue 3
Identifying boosted objects with N-subjettiness
journal, March 2011
- Thaler, Jesse; Van Tilburg, Ken
- Journal of High Energy Physics, Vol. 2011, Issue 3
Jet flavor classification in high-energy physics with deep neural networks
journal, December 2016
- Guest, Daniel; Collado, Julian; Baldi, Pierre
- Physical Review D, Vol. 94, Issue 11
Learning Particle Physics by Example: Location-Aware Generative Adversarial Networks for Physics Synthesis
journal, September 2017
- de Oliveira, Luke; Paganini, Michela; Nachman, Benjamin
- Computing and Software for Big Science, Vol. 1, Issue 1
Accelerating Science with Generative Adversarial Networks: An Application to 3D Particle Showers in Multilayer Calorimeters
journal, January 2018
- Paganini, Michela; de Oliveira, Luke; Nachman, Benjamin
- Physical Review Letters, Vol. 120, Issue 4
Long Short-Term Memory
journal, November 1997
- Hochreiter, Sepp; Schmidhuber, Jürgen
- Neural Computation, Vol. 9, Issue 8
Hard-soft-collinear factorization to all orders
journal, November 2014
- Feige, Ilya; Schwartz, Matthew D.
- Physical Review D, Vol. 90, Issue 10
Successive combination jet algorithm for hadron collisions
journal, October 1993
- Ellis, Stephen D.; Soper, Davison E.
- Physical Review D, Vol. 48, Issue 7
Better jet clustering algorithms
journal, August 1997
- Dokshitzer, Yu. L.; Leder, G. D.; Moretti, S.
- Journal of High Energy Physics, Vol. 1997, Issue 08