Graph Neural Networks in Particle Physics

Shlomi, Jonathan; Battaglia, Peter; Vlimant, Jean-Roch

doi:10.48550/arxiv.2007.13681

Cited by 11 publications

(11 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…GNNs [59][60][61] have been studied for jet classification in supervised [57,58,[62][63][64][65] as well as unsupervised [66,67] scenarios and has state-of-the-art performance [57] compared to other still excellent architectures [68] like Convolutional Neural Networks (CNNs), Deep-sets, and Recurrent Neural Networks (RNNs). The better performance originates in GNNs having an inductive bias more appropriate for jet substructure in particular and collider physics in general.…”

Section: Jhep02(2022)060mentioning

confidence: 99%

Energy-weighted message passing: an infra-red and collinear safe graph neural network algorithm

Konar

Ngairangbam

Spannowsky

2022

J. High Energ. Phys.

View full text Add to dashboard Cite

Hadronic signals of new-physics origin at the Large Hadron Collider can remain hidden within the copiously produced hadronic jets. Unveiling such signatures require highly performant deep-learning algorithms. We construct a class of Graph Neural Networks (GNN) in the message-passing formalism that makes the network output infra-red and collinear (IRC) safe, an important criterion satisfied within perturbative QCD calculations. Including IRC safety of the network output as a requirement in the construction of the GNN improves its explainability and robustness against theoretical uncertainties in the data. We generalise Energy Flow Networks (EFN), an IRC safe deep-learning algorithm on a point cloud, defining energy weighted local and global readouts on GNNs. Applying the simplest of such networks to identify top quarks, W bosons and quark/gluon jets, we find that it outperforms state-of-the-art EFNs. Additionally, we obtain a general class of graph construction algorithms that give structurally invariant graphs in the IRC limit, a necessary criterion for the IRC safety of the GNN output.

show abstract

Section: Jhep02(2022)060mentioning

confidence: 99%

Energy-weighted message passing: an infra-red and collinear safe graph neural network algorithm

Konar

Ngairangbam

Spannowsky

2022

J. High Energ. Phys.

View full text Add to dashboard Cite

show abstract

“…A few examples of such applications are jet-tagging [7,8], secondary vertex finding [9], event reconstruction [10][11][12][13], and jet parton assignment [14]. A comprehensive review of the different methods is described in [15].…”

Section: Related Workmentioning

confidence: 99%

Point Cloud Transformers applied to Collider Physics

Mikuni,

Canelli

2021

Preprint

View full text Add to dashboard Cite

Methods for processing point cloud information have seen a great success in collider physics applications. One recent breakthrough in machine learning is the usage of Transformer networks to learn semantic relationships between sequences in language processing. In this work, we apply a modified Transformer network called Point Cloud Transformer as a method to incorporate the advantages of the Transformer architecture to an unordered set of particles resulting from collision events. To compare the performance with other strategies, we study jet-tagging applications for highly-boosted particles.

show abstract

“…Recent work suggests, that deep learning models can be an efficient and accurate alternative to traditional, hand-crafted simulation approaches [18,13,21]. Graph networks in particular have become popular in collider physics [20], astrophysics [6] or chemistry [12] and have been used to build various data-driven simulators [4,12,3,19,16].…”

Section: Introductionmentioning

confidence: 99%

Simulating Liquids with Graph Networks

Klimesch¹,

Holl²,

Thuerey³

2022

Preprint

View full text Add to dashboard Cite

Simulating complex dynamics like fluids with traditional simulators is computationally challenging. Deep learning models have been proposed as an efficient alternative, extending or replacing parts of traditional simulators. We investigate graph neural networks (GNNs) for learning fluid dynamics and find that their generalization capability is more limited than previous works would suggest. We also challenge the current practice of adding random noise to the network inputs in order to improve its generalization capability and simulation stability. We find that inserting the real data distribution, e.g. by unrolling multiple simulation steps, improves accuracy and that hiding all domain-specific features from the learning model improves generalization. Our results indicate that learning models, such as GNNs, fail to learn the exact underlying dynamics unless the training set is devoid of any other problem-specific correlations that could be used as shortcuts.

show abstract

Graph Neural Networks in Particle Physics

Cited by 11 publications

References 0 publications

Energy-weighted message passing: an infra-red and collinear safe graph neural network algorithm

Energy-weighted message passing: an infra-red and collinear safe graph neural network algorithm

Point Cloud Transformers applied to Collider Physics

Simulating Liquids with Graph Networks

Contact Info

Product

Resources

About