The Machine Learning landscape of top taggers

@article{Kasieczka2019TheML,
  title={The Machine Learning landscape of top taggers},
  author={Gregor Kasieczka and Tilman Plehn and Anja Butter and Dipsikha Debnath and M. Fairbairn and Wojciech Fedorko and Colin Gay and Loukas Gouskos and Patrick T. Komiske and S. Leiss and Alison Lister and Sebastian Macaluso and Eric M. Metodiev and Liam Moore and Benjamin Philip Nachman and Karl Nordstrom and Jannicke Pearkes and Huilin Qu and Yannik Rath and M A Riegler and David Shih and J. Thompson and Sreedevi Varma},
  journal={SciPost Physics},
  year={2019}
}
Based on the established task of identifying boosted, hadronically decaying top quarks, we compare a wide range of modern machine learning approaches. Unlike most established methods they rely on low-level input, for instance calorimeter output. While their network architectures are vastly different, their performance is comparatively similar. In general, we find that these new approaches are extremely powerful and great fun. 

Figures and Tables from this paper

Performance of fully-connected neural networks for top tagging
TLDR
Although fully-connected neural networks have the simplest architecture, they still have a good performance for identifying boosted top quarks and could reach accuracy of about 89%.
Boosting mono-jet searches with model-agnostic machine learning
TLDR
It is shown how weakly supervised machine learning can improve the sensitivity of LHC mono-jet searches to new physics models with anomalous jet dynamics and the discovery potential of an existing generic search can be boosted considerably.
Boosted top quark tagging and polarization measurement using machine learning
Machine learning techniques are used for treating jets as images to explore the performance of boosted top quark tagging. Tagging performances are studied in both hadronic and leptonic channels of
Towards machine learning analytics for jet substructure
TLDR
The aim of this paper is to move a first step into the direction of applying expert-knowledge in particle physics to calculate the optimal decision function and test whether it is achieved by standard training, thus making the aforementioned black-box more transparent.
Deep-learning jets with uncertainties and more
TLDR
The main features of Bayesian versions of established deep-learning taggers are illustrated and it is shown how they capture statistical uncertainties from finite training samples, systematics related to the jet energy scale, and stability issues through pile-up.
Machine Learning Scientific Competitions and Datasets
TLDR
Four scientific competitions with the objective of discovering innovative techniques to perform typical High Energy Physics tasks, like event reconstruction, classification and new physics discovery are summarised in this chapter.
Boosted Top Quark Tagging and Polarization 2 Measurement using Machine Learning
Machine learning techniques are used to explore the performance of boosted top quark tagging treating jets as images. Tagging performances are studied in both hadronic and leptonic channels employing
PyTorch Neural Networks and Track Analysis for Top Quark Tagging
TLDR
Deep Neural Networks and Long Short-Term Memory networks were built in PyTorch to compare their performances to previously tested Keras models and it was found that keeping 100 pT ordered tracks with pT greater than 1 GeV could retain relevant information for jet classification while minimizing noise and computing time.
A Living Review of Machine Learning for Particle Physics
TLDR
This living review is a nearly comprehensive list of citations for those developing and applying deep learning approaches to experimental, phenomenological, or theoretical analyses, and will be updated as often as possible to incorporate the latest developments.
Mass agnostic jet taggers
TLDR
This is the first study to show that the data augmentation techniques of Planing and PCA based scaling deliver similar performance as the augmented training techniques of Adversarial NN and uBoost, but are both easier to implement and computationally cheaper.
...
...

References

SHOWING 1-10 OF 123 REFERENCES
Deep-learned Top Tagging with a Lorentz Layer
TLDR
A new and highly efficient tagger for hadronically decaying topquarks is introduced, based on a deep neural network working with Lorentz vectors and the Minkowski metric, which significantly increases the performance for strongly boosted top quarks.
Deep-learning top taggers or the end of QCD?
A bstractMachine learning based on convolutional neural networks can be used to study jet images from the LHC. Top tagging in fat jets offers a well-defined framework to establish our DeepTop
Jet Constituents for Deep Neural Network Based Top Quark Tagging
TLDR
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 to multiple proton-proton interactions at the levels expected throughout Run 2 of the LHC.
Parton Shower Uncertainties in Jet Substructure Analyses with Deep Neural Networks
Machine learning methods incorporating deep neural networks have been the subject of recent proposals for new hadronic resonance taggers. These methods require training on a data set produced by an
Infrared safety of a neural-net top tagging algorithm
TLDR
This paper constructs a top-jet tagger based on a Convolutional Neural Network, and applies it to parton-level boosted top samples, with and without an additional gluon in the final state, showing that the jet observable defined by the CNN obeys the canonical definition of infrared safety.
QCD or what?
TLDR
This work shows how adversarial autoencoder networks, trained only on QCD jets, identify jets from decays of arbitrary heavy resonances using 4-vectors, allowing for a general and at the same time easily controllable search for new physics.
Playing tag with ANN: boosted top identification with pattern recognition
A bstractMany searches for physics beyond the Standard Model at the Large Hadron Collider (LHC) rely on top tagging algorithms, which discriminate between boosted hadronic top quarks and the much
Weakly supervised classification in high energy physics
TLDR
Using one of the most challenging binary classification tasks in high energy physics — quark versus gluon tagging — it is shown that weakly supervised classification can match the performance of fully supervised algorithms.
Performance of top-quark and $$\varvec{W}$$ W -boson tagging with ATLAS in Run 2
The performance of identification algorithms (“taggers”) for hadronically decaying top quarks and W bosons in pp collisions at √ s = 13 TeV recorded by the ATLAS experiment at the Large Hadron
Jet-images — deep learning edition
TLDR
This interplay between physicallymotivated feature driven tools and supervised learning algorithms is general and can be used to significantly increase the sensitivity to discover new particles and new forces, and gain a deeper understanding of the physics within jets.
...
...