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

Konar, Partha; Ngairangbam, Vishal S.; Spannowsky, Michael

Abstract

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.

Citation

Konar, P., Ngairangbam, V. S., & Spannowsky, M. (2022). Energy-weighted message passing: an infra-red and collinear safe graph neural network algorithm. Journal of High Energy Physics, 2022(2), Article 60. https://doi.org/10.1007/jhep02%282022%29060

• Journal Article Type: Article
• Acceptance Date: Jan 25, 2022
• Online Publication Date: Feb 8, 2022
• Publication Date: 2022
• Deposit Date: May 4, 2022
• Publicly Available Date: May 5, 2022
• Journal: Journal of High Energy Physics
• Print ISSN: 1126-6708
• Electronic ISSN: 1029-8479
• Publisher: Scuola Internazionale Superiore di Studi Avanzati (SISSA)
• Peer Reviewed: Peer Reviewed
• Volume: 2022
• Issue: 2
• Article Number: 60
• DOI: https://doi.org/10.1007/jhep02%282022%29060
• Public URL: https://durham-repository.worktribe.com/output/1208862