Graph Neural Networks in Particle Physics: Implementations, Innovations, and Challenges

Thais, S. J.; Calafiura, P.; Chachamis, Grigorios; Dezoort, G.; Duarte, J.; Ganguly, S.; Kagan, M.; Murnane, Daniel; Neubauer, M. S.; Terao, K.

doi:10.48550/arxiv.2203.12852

Cited by 14 publications

(13 citation statements)

References 100 publications

(119 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Furthermore, most of the voxels in each sample are empty, leading to sparse data, which CNNs were not originally designed for. In this context, we intend to apply Graph Neural Networks (GNNs) as the base model, which have shown success in reconstruction tasks for neutrino physics applications (Ju et al 2020;Thais et al 2022). With respect to the uncertainty estimation, the absence of the off-diagonal terms in the predicted covariance between the different spatial directions likely led to a tendency towards larger prediction intervals than would have existed if off-diagonal terms were included.…”

Section: Discussionmentioning

confidence: 99%

Low-Energy Electron-Track Imaging for a Liquid Argon Time-Projection-Chamber Telescope Concept using Probabilistic Deep Learning

Buuck¹,

Mishra²,

Charles³

et al. 2022

Preprint

View full text Add to dashboard Cite

The GammaTPC is an MeV-scale single-phase liquid argon time-projection-chamber gamma-ray telescope concept with a novel dual-scale pixel-based charge-readout system. It promises to enable a significant improvement in sensitivity to MeVscale gamma-rays over previous telescopes. The novel pixel-based charge readout allows for imaging of the tracks of electrons scattered by Compton interactions of incident gamma-rays. The two primary contributors to the accuracy of a Compton telescope in reconstructing an incident gamma-ray's original direction are its energy and position resolution. In this work, we focus on using deep learning to optimize the reconstruction of the initial position and direction of electrons scattered in Compton interactions, including using probabilistic models to estimate predictive uncertainty. We show that the deep learning models are able to predict locations of Compton scatters of MeV-scale gamma-rays from simulated pixel-based data to better than 0.6 mm RMS error, and are sensitive to the initial direction of the scattered electron. We compare and contrast different deep learning uncertainty estimation algorithms for reconstruction applications. Additionally, we show that event-by-event estimates of the uncertainty of the locations of the Compton scatters can be used to select those events that were reconstructed most accurately, leading to improvement in locating the origin of gamma-ray sources on the sky.

show abstract

Section: Discussionmentioning

confidence: 99%

Low-Energy Electron-Track Imaging for a Liquid Argon Time-Projection-Chamber Telescope Concept using Probabilistic Deep Learning

Buuck¹,

Mishra²,

Charles³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…The use of neural networks for end-to-end reconstruction goes beyond calorimeter reconstruction or tracking. Computer vision techniques based on convolutional and graph neural networks have been used for event-topology classification directly from the energy map of detector hits [28][29][30][31]. Deep neural networks have been exploited as a tool to cluster high-level objects, e.g., particle-flow candidates [32] and jets [33,34].…”

Section: Related Workmentioning

confidence: 99%

End-to-end multi-particle reconstruction in high occupancy imaging calorimeters with graph neural networks

et al. 2022

View full text Add to dashboard Cite

We present an end-to-end reconstruction algorithm to build particle candidates from detector hits in next-generation granular calorimeters similar to that foreseen for the high-luminosity upgrade of the CMS detector. The algorithm exploits a distance-weighted graph neural network, trained with object condensation, a graph segmentation technique. Through a single-shot approach, the reconstruction task is paired with energy regression. We describe the reconstruction performance in terms of efficiency as well as in terms of energy resolution. In addition, we show the jet reconstruction performance of our method and discuss its inference computational cost. To our knowledge, this work is the first-ever example of single-shot calorimetric reconstruction of $${\mathcal {O}}(1000)$$ O ( 1000 ) particles in high-luminosity conditions with 200 pileup.

show abstract

“…We considered several options to apply Machine Learning approach to this problem. First of all, the definition of the jets assignment problem is similar to reconstruction (clustering) tasks [46][47][48][49][50] where Graph Neural Networks (GNN) inspired architectures were applied. The target of GNN is to establish edges or connections between input points.…”

Section: Detector-levelmentioning

confidence: 99%

Prospects for Higgs boson and new scalar resonant production searches in ttbb final state at the LHC

Mandrik¹

2022

Preprint

View full text Add to dashboard Cite

In this article we probe resonant associated production of a Standard Model Higgs boson with new heavy scalar resonance in proton-proton collisions at a center-of-mass energy √ s = 13 TeV. The Higgs boson and new scalar resonant are required to decay into a pair of bottom quarks and a pair of top quarks, respectively. Semileptonic decay of top quarks is considered. The searches are projected into operation conditions of the Large Hadron Collider during Run II data taking period at a center-of-mass energy of 13 TeV using Monte Carlo generated events, realistic detector response simulation and available Open Data samples. Analysis strategies are presented and machine learning approach using Deep Neural Network is proposed to resolve ambiguous in jets assignment and improve kinematic reconstruction of signal events. Sensitivity of the CMS detector is estimated as 95% expected upper limits on the product of the production cross section and the branching fractions of the searched particles.

show abstract

Graph Neural Networks in Particle Physics: Implementations, Innovations, and Challenges

Cited by 14 publications

References 100 publications

Low-Energy Electron-Track Imaging for a Liquid Argon Time-Projection-Chamber Telescope Concept using Probabilistic Deep Learning

Low-Energy Electron-Track Imaging for a Liquid Argon Time-Projection-Chamber Telescope Concept using Probabilistic Deep Learning

End-to-end multi-particle reconstruction in high occupancy imaging calorimeters with graph neural networks

Prospects for Higgs boson and new scalar resonant production searches in ttbb final state at the LHC

Contact Info

Product

Resources

About