319 Hits in 10.8 sec

Improving robustness of jet tagging algorithms with adversarial training [article]

Annika Stein and Xavier Coubez and Spandan Mondal and Andrzej Novak and Alexander Schmidt
2022 arXiv   pre-print
Subsequently, we present an adversarial training strategy that mitigates the impact of such simulated attacks and improves the classifier robustness.  ...  We investigate the classifier response to input data with injected mismodelings and probe the vulnerability of flavor tagging algorithms via application of adversarial attacks.  ...  We thank Nicolas Frediani for his contributions to the project in context of his Bachelor thesis. Compliance with ethical standards  ... 
arXiv:2203.13890v1 fatcat:h3gwt5537nfongsj4c57xkuq3e

Interaction networks for the identification of boosted H→bb¯ decays

Eric A. Moreno, Thong Q. Nguyen, Jean-Roch Vlimant, Olmo Cerri, Harvey B. Newman, Avikar Periwal, Maria Spiropulu, Javier M. Duarte, Maurizio Pierini
2020 Physical Review D  
The interaction network achieves a drastic improvement in the identification performance with respect to state-of-the-art algorithms.  ...  The algorithm is trained on simulated samples of realistic LHC collisions, released by the CMS Collaboration on the CERN Open Data Portal.  ...  Adversarial training The secondary adversary network is constructed that consists of three hidden layers each with 64 nodes.  ... 
doi:10.1103/physrevd.102.012010 fatcat:qbny2obwqjarhjqxhkkgvduety

Interaction networks for the identification of boosted H→ bb decays [article]

Eric A. Moreno and Thong Q. Nguyen and Jean-Roch Vlimant and Olmo Cerri and Harvey B. Newman and Avikar Periwal and Maria Spiropulu and Javier M. Duarte and Maurizio Pierini
2019 arXiv   pre-print
The interaction network achieves a drastic improvement in the identification performance with respect to state-of-the-art algorithms.  ...  The algorithm is trained on simulated samples of accurate LHC collisions, released by the CMS collaboration on the CERN Open Data Portal.  ...  The training begins by initializing the weights from the best classifier training. The adversary is then pre-trained for 10 epochs using the Adam algorithm with an initial learning rate of 10 −4 .  ... 
arXiv:1909.12285v3 fatcat:gbrgymajbnhuhn26hwdlezjscy

Robust Jet Classifiers through Distance Correlation

Gregor Kasieczka, David Shih
2020 Physical Review Letters  
We also show the feasibility of regularization with distance correlation for more powerful convolutional neural networks, as well as for the problem of hadronic top tagging.  ...  networks but is much simpler and more stable to train.  ...  We are grateful to Chris Delitzsch, Steven Schramm, and especially Andreas Sogaard for help with details of the ATLAS decorrelation study. G.  ... 
doi:10.1103/physrevlett.125.122001 pmid:33016758 fatcat:fxx54bv3qfgm5bxhfyq4d2mu3u

DisCo Fever: Robust Networks Through Distance Correlation [article]

Gregor Kasieczka, David Shih
2020 arXiv   pre-print
networks but is much simpler to train and has better convergence properties.  ...  We also show the feasibility of DisCo regularization for more powerful convolutional neural networks, as well as for the problem of hadronic top tagging.  ...  Training with adversary Adversarial training follows the approach outlined in the Introduction, with the adversary attempting to learn the PDF of the mass.  ... 
arXiv:2001.05310v1 fatcat:obfin7eadvhspictn2bdtkwgia

Decorrelated jet substructure tagging using adversarial neural networks

Chase Shimmin, Peter Sadowski, Pierre Baldi, Edison Weik, Daniel Whiteson, Edward Goul, Andreas Søgaard
2017 Physical Review D  
We generalize the adversarial training technique to include a parametric dependence on the signal hypothesis, training a single network that provides optimized, interpolatable decorrelated jet tagging  ...  The network is trained using an adversarial strategy, resulting in a tagger that learns to balance classification accuracy with decorrelation.  ...  The authors acknowledge useful conversations with Kyle Cranmer, Jesse Thaler, Kevin Bauer, and Dan Guest, helpful comments from Sal Rappoccio, Derek Soeder and Michela Paganini, and are grateful to the  ... 
doi:10.1103/physrevd.96.074034 fatcat:odefx5vyvbfete3llbwwh2m2a4

Adversarially-trained autoencoders for robust unsupervised new physics searches

Andrew Blance, Michael Spannowsky, Philip Waite
2019 Journal of High Energy Physics  
To mitigate their effect on the classifier and to allow for a realistic assessment of the method, we propose to combine the autoencoder with an adversarial neural network to remove its sensitivity to the  ...  We quantify its effect and show that one can achieve a robust anomaly detection in resonance-induced tt̅ final states.  ...  Acknowledgments MS acknowledges the generous hospitality of Barbara Jaeger and her group at the University of Tuebingen, as well as support of the Humboldt Society, during the completion of parts of this  ... 
doi:10.1007/jhep10(2019)047 fatcat:gmrqoxldpbbo3aavjampmwczay

Jet Constituents for Deep Neural Network Based Top Quark Tagging [article]

Jannicke Pearkes, Wojciech Fedorko, Alison Lister, Colin Gay
2017 arXiv   pre-print
Here, a sequential approach to this task is taken by using an ordered sequence of jet constituents as training inputs.  ...  The jet classification method achieves a background rejection of 45 at a 50% efficiency operating point for reconstruction level jets with transverse momentum range of 600 to 2500 GeV and is insensitive  ...  [21, 28, 30, 31] have explored the performance of the algorithms top tagging.  ... 
arXiv:1704.02124v2 fatcat:lgeuibwj25bdvp3wkhsufwfdwq

Meta-learning and data augmentation for mass-generalised jet taggers [article]

Matthew J. Dolan, Ayodele Ore
2021 arXiv   pre-print
Deep neural networks trained for jet tagging are typically specific to a narrow range of transverse momenta or jet masses.  ...  The meta-learning algorithms provide only a small improvement in generalisation when combined with this augmentation.  ...  Computing resources were provided by the LIEF HPC-GPGPU Facility hosted at the University of Melbourne. This Facility was established with the assistance of LIEF Grant LE170100200.  ... 
arXiv:2111.06047v2 fatcat:vicbebenirblhgybsp7mwrnujy

Deep Learning and Its Application to LHC Physics

Dan Guest, Kyle Cranmer, Daniel Whiteson
2018 Annual Review of Nuclear and Particle Science  
This review is aimed at the reader who is familiar with high energy physics but not machine learning.  ...  the power of deep learning for analysis of LHC data, and discussion of future prospects and concerns.  ...  D.W. and D.G. are supported by the Office of Science at the US Department of Energy.  ... 
doi:10.1146/annurev-nucl-101917-021019 fatcat:4ll2ex624jcutgimi5w7wya2bq

Image-Based Jet Analysis [article]

Michael Kagan
2020 arXiv   pre-print
Image-based jet analysis is built upon the jet image representation of jets that enables a direct connection between high energy physics and the fields of computer vision and deep learning.  ...  Through this connection, a wide array of new jet analysis techniques have emerged.  ...  (b) ROC curves for quark / gluon jet rejection versus top jet efficiency for jet image based adversarial autoencoders with varying strength of adversarial penalty during training [79] .  ... 
arXiv:2012.09719v2 fatcat:lkmk52e6gvfevakbvsp2bmavl4

Achieving Model Robustness through Discrete Adversarial Training [article]

Maor Ivgi, Jonathan Berant
2021 arXiv   pre-print
Furthermore, online augmentation with search-based attacks justifies the higher training cost, significantly improving robustness on three datasets.  ...  While such attacks have been extensively explored for the purpose of evaluating model robustness, their utility for improving robustness has been limited to offline augmentation only.  ...  Our online augmentation of adversarial examples (ADVON, yellow circles) significantly improves robust accuracy, but is expensive to train.  ... 
arXiv:2104.05062v2 fatcat:dh63egxs6fc6bg7zjnpwm47v4y

Fast and accurate simulation of particle detectors using generative adversarial networks [article]

Pasquale Musella, Francesco Pandolfi
2018 arXiv   pre-print
We show that deep neural networks can achieve high-fidelity in this task, while attaining a speed increase of several orders of magnitude with respect to traditional algorithms.  ...  In this work we apply this kind of technique to the simulation of particle-detector response to hadronic jets.  ...  We strongly support this initiative and believe that it will be crucial to spark developments of new algorithms from which the HEP community as a whole can profit. We thank Dr. M. DonegÃă, Prof. G.  ... 
arXiv:1805.00850v1 fatcat:z7lwjj4cw5aldip2253sf765vy

AI Safety for High Energy Physics [article]

Benjamin Nachman, Chase Shimmin
2019 arXiv   pre-print
In addition to providing a pragmatic diagnostic, this work will hopefully begin a dialogue within the community about the robust application of deep learning to experimental analyses.  ...  The field of high-energy physics (HEP), along with many scientific disciplines, is currently experiencing a dramatic influx of new methodologies powered by modern machine learning techniques.  ...  ACKNOWLEDGMENTS We thank Paul Tipton for constructive feedback since the early stages of this project. We are also grate-  ... 
arXiv:1910.08606v1 fatcat:qdlnlfmpqjay7nuzn5arlw3gk4

Classification without labels: learning from mixed samples in high energy physics

Eric M. Metodiev, Benjamin Nachman, Jesse Thaler
2017 Journal of High Energy Physics  
training samples.  ...  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.  ...  This article is distributed under the terms of the Creative Commons Attribution License (CC-BY 4.0), which permits any use, distribution and reproduction in any medium, provided the original author(s)  ... 
doi:10.1007/jhep10(2017)174 fatcat:cukk3mxxgfckxdsgjlehbp5rtq
« Previous Showing results 1 — 15 out of 319 results