Classification without labels: learning from mixed samples in high energy physics
Reads0
Chats0
TLDR
In this paper, classification without labels (CWoLa) is proposed to distinguish statistical mixtures of classes, which are common in collider physics, where neither individual labels nor class proportions are required, yet they prove that the optimal classifier in the CWoLa paradigm is also the optimal one in the traditional fully supervised case where all label information is available.Abstract:
Modern machine learning techniques can be used to construct powerful models for difficult collider physics problems. In many applications, however, these models are trained on imperfect simulations due to a lack of truth-level information in the data, which risks the model learning artifacts of the simulation. In this paper, we introduce the paradigm of classification without labels (CWoLa) in which a classifier is trained to distinguish statistical mixtures of classes, which are common in collider physics. Crucially, neither individual labels nor class proportions are required, yet we prove that the optimal classifier in the CWoLa paradigm is also the optimal classifier in the traditional fully-supervised case where all label information is available. After demonstrating the power of this method in an analytical toy example, we consider a realistic benchmark for collider physics: distinguishing quark- versus gluon-initiated jets using mixed quark/gluon training samples. More generally, CWoLa can be applied to any classification problem where labels or class proportions are unknown or simulations are unreliable, but statistical mixtures of the classes are available.read more
Citations
More filters
Journal ArticleDOI
Machine learning and the physical sciences
Giuseppe Carleo,J. Ignacio Cirac,Kyle Cranmer,Laurent Daudet,Maria Schuld,Naftali Tishby,Leslie Vogt-Maranto,Lenka Zdeborová +7 more
TL;DR: This article reviews in a selective way the recent research on the interface between machine learning and the physical sciences, including conceptual developments in ML motivated by physical insights, applications of machine learning techniques to several domains in physics, and cross fertilization between the two fields.
Journal ArticleDOI
Jet Substructure at the Large Hadron Collider: A Review of Recent Advances in Theory and Machine Learning
TL;DR: A comprehensive review of state-of-the-art theoretical and machine learning developments in jet substructure is provided in this article, which is meant both as a pedagogical introduction and as a comprehensive reference for experts.
Journal ArticleDOI
Jet tagging via particle clouds
Huilin Qu,Loukas Gouskos +1 more
TL;DR: This work proposes ParticleNet, a customized neural network architecture using Dynamic Graph Convolutional Neural Network for jet tagging problems that achieves state-of-the-art performance on two representative jet tagging benchmarks and is improved significantly over existing methods.
Journal ArticleDOI
Searching for long-lived particles beyond the Standard Model at the Large Hadron Collider
Juliette Alimena,James Baker Beacham,Martino Borsato,Yangyang Cheng,Xabier Cid Vidal,Giovanna Cottin,Giovanna Cottin,Giovanna Cottin,David Curtin,Albert De Roeck,Nishita Desai,J. Evans,Simon Knapen,Sabine Kraml,Andre Lessa,Zhen Liu,Sascha Mehlhase,M. Ramsey-Musolf,M. Ramsey-Musolf,Heather Lynn Russell,Jessie Shelton,Brian Shuve,Brian Shuve,Monica Verducci,José Zurita,Todd Adams,Michael Adersberger,Cristiano Alpigiani,Artur Apresyan,Robert Bainbridge,Varvara Batozskaya,Hugues Beauchesne,Lisa Benato,Simon Berlendis,Eshwen Bhal,Freya Blekman,Christina Borovilou,Jamie Boyd,Benjamin Brau,Lene Bryngemark,Oliver Buchmueller,Malte Buschmann,William Buttinger,Mario Campanelli,Cari Cesarotti,Chunhui Chen,Hsin-Chia Cheng,Sanha Cheong,Matthew Citron,Andrea Coccaro,Victor Coco,Eric Conte,Felix Cormier,Louie Corpe,Nathaniel Craig,Y. Cui,Elena Dall'Occo,C. Dallapiccola,Mohamed Rashad Darwish,Alessandro Davoli,Annapaola De Cosa,Andrea De Simone,Luigi Delle Rose,Luigi Delle Rose,Frank F. Deppisch,Biplab Dey,Miriam Diamond,Keith R. Dienes,Keith R. Dienes,Sven Dildick,Babette Döbrich,Marco Drewes,Melanie Eich,M. Elsawy,M. Elsawy,Alberto Escalante Del Valle,Gabriel John Facini,Marco Farina,Jonathan L. Feng,Oliver Fischer,Henning Flaecher,Patrick Foldenauer,Marat Freytsis,Marat Freytsis,Benjamin Fuks,Iftah Galon,Yuri Gershtein,Stefano Giagu,Andrea Giammanco,Vladimir Gligorov,Tobias Golling,Sergio Grancagnolo,Giuliano Gustavino,Andy Haas,Kristian Hahn,Jan Hajer,Ahmed Hammad,Lukas Heinrich,Jan Heisig,Juan Carlos Helo,Gavin Grant Hesketh,Christopher S. Hill,Martin Hirsch,Marcus Hohlmann,Tova Ray Holmes,Wouter Hulsbergen,John Huth,Philip Ilten,Thomas Jacques,Bodhitha Jayatilaka,Geng Yuan Jeng,K. A. Johns,Toshiaki Kaji,Gregor Kasieczka,Yevgeny Kats,Malgorzata Kazana,Henning Keller,Maxim Yu. Khlopov,Felix Kling,Ted Kolberg,Igor Kostiuk,Emma Sian Kuwertz,Audrey Katherine Kvam,Greg Landsberg,Gaia Lanfranchi,Inaki Lara,Alexander Ledovskoy,Dylan Linthorne,Jia Liu,Iacopo Longarini,Steven Lowette,Henry Lubatti,Margaret Susan Lutz,Jingyu Luo,Judita Mamuzic,Matthieu Marinangeli,Alberto Mariotti,Daniel Robert Marlow,Matthew McCullough,Kevin Mcdermott,Philippe Mermod,David Milstead,Siddharth Mishra-Sharma,Vasiliki A Mitsou,Javier Montejo Berlingen,Filip Moortgat,Filip Moortgat,Alessandro Morandini,Alice Polyxeni Morris,David Michael Morse,Stephen Mrenna,Benjamin Philip Nachman,Miha Nemevšek,Fabrizio Nesti,Christian Ohm,Christian Ohm,Silvia Pascoli,Kevin Pedro,Cristian Pena,Cristian Pena,Karla Josefina Pena Rodriguez,Jónatan Piedra,James Pinfold,Antonio Policicchio,Goran Popara,Jessica Prisciandaro,Mason Proffitt,Giorgia Rauco,F. Redi,Matthew Reece,Allison Reinsvold Hall,H. Rejeb Sfar,Sophie Renner,Dean J. Robinson,Dean J. Robinson,Amber Roepe,Manfredi Ronzani,Ennio Salvioni,Arka Santra,Ryu Sawada,Jakub Scholtz,Philip Schuster,Pedro Schwaller,Cristiano David Sebastiani,Sezen Sekmen,Michele Selvaggi,Weinan Si,Livia Soffi,Daniel Stolarski,David Stuart,John Stupak,Kevin Sung,Wendy Taylor,Sebastian Templ,Brooks Thomas,Emma Torró-Pastor,Daniele Trocino,Sebastian Trojanowski,Marco Trovato,Yuhsin Tsai,Christopher George Tully,Tamás Álmos Vámi,Juan Carlos Vasquez,Carlos Vázquez Sierra,K. Vellidis,Basile Vermassen,Martina Vit,Devin G. E. Walker,Xiaoping Wang,Gordon Watts,Si Xie,Melissa Yexley,C. C. Young,Jiang Hao Yu,Jiang Hao Yu,Piotr Zalewski,Yongchao Zhang +216 more
TL;DR: In this paper, the authors present a survey of the current state of LLP searches at the Large Hadron Collider (LHC) and chart a path for the development of LLP searches into the future, both in the upcoming Run 3 and at the high-luminosity LHC.
Journal ArticleDOI
Anomaly Detection with Density Estimation
TL;DR: It is shown how ANODE can enhance the significance of a dijet bump hunt by up to a factor of 7 with a 10\% accuracy on the background prediction, and is robust against systematic differences between signal region and sidebands, giving it broader applicability than other methods.
References
More filters
Proceedings ArticleDOI
Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification
TL;DR: In this paper, a Parametric Rectified Linear Unit (PReLU) was proposed to improve model fitting with nearly zero extra computational cost and little overfitting risk, which achieved a 4.94% top-5 test error on ImageNet 2012 classification dataset.
Journal ArticleDOI
The anti-$k_t$ jet clustering algorithm
TL;DR: The anti-k-t algorithm as mentioned in this paper behaves like an idealised cone algorithm, in that jets with only soft fragmentation are conical, active and passive areas are equal, the area anomalous dimensions are zero, the non-global logarithms are those of a rigid boundary and the Milan factor is universal.
Posted Content
TensorFlow: A system for large-scale machine learning
Martín Abadi,Paul Barham,Jianmin Chen,Zhifeng Chen,Andy Davis,Jeffrey Dean,Matthieu Devin,Sanjay Ghemawat,Geoffrey Irving,Michael Isard,Manjunath Kudlur,Josh Levenberg,Rajat Monga,Sherry Moore,Derek G. Murray,Benoit Steiner,Paul A. Tucker,Vijay K. Vasudevan,Pete Warden,Martin Wicke,Yuan Yu,Xiaoqiang Zheng +21 more
TL;DR: The TensorFlow dataflow model is described and the compelling performance that Tensor Flow achieves for several real-world applications is demonstrated.
Journal ArticleDOI
A Brief Introduction to PYTHIA 8.1
TL;DR: PYTHIA 8 represents a complete rewrite in C++, and does not yet in every respect replace the old code, but does contain some new physics aspects that should make it an attractive option especially for LHC physics studies.
Journal ArticleDOI
FastJet User Manual
TL;DR: FastJet as mentioned in this paper is a C++ package that provides a broad range of jet finding and analysis tools, including efficient native implementations of all widely used 2→1 sequential recombination jet algorithms for pp and e − − collisions.