Supervised jet clustering with graph neural networks for Lorentz boosted bosons

Abstract

Jet clustering is traditionally an unsupervised learning task because there is no unique way to associate hadronic final states with the quark and gluon degrees of freedom that generated them. However, for uncolored particles like W, Z, and Higgs bosons, it is possible to approximately (though not exactly) associate final state hadrons to their ancestor. By labeling simulated final state hadrons as descending from an uncolored particle, it is possible to train a supervised learning method to create boson jets. Such a method would operate on individual particles and identify connections between particles originating from the same uncolored particle. Graph neural networks are well-suited for this purpose as they can act on unordered sets and naturally create strong connections between particles with the same label. These networks are used to train a supervised jet clustering algorithm. The kinematic properties of these graph jets better match the properties of simulated Lorentz-boosted W bosons. Furthermore, the graph jets contain more information for discriminating W jets from generic quark jets. This work marks the beginning of a new exploration in jet physics to use machine learning to optimize the construction of jets and not only the observables computed from jet constituents.

Reconstructing boosted Higgs jets from event image segmentation

“…Note Added. While we were preparing this manuscript, a paper with similar purpose [74] appears on the arXiv. Different neural network was adopted in their study.…”

Higgs and top physics reconstruction challenges and opportunities at FCC-ee

“…This area has not yet been thoroughly explored, and may hold significant potential for further improvement with more advanced analysis strategies. For Higgs boson physics, strategies for "colour singlet clustering" [31][32][33][34], that exploit the physical properties of colour-neutral particles decaying into two hadronic jets, may improve the matching of particles to jets, resulting in a better measurement of the properties of the underlying original partons. Such techniques, if proven to be capable of improved precision for the reconstruction of signal final states while also being robust against backgrounds, would improve the measurement of the global event kinematics, and would result in a clearer separation of signal and background.…”

MLPF: efficient machine-learned particle-flow reconstruction using graph neural networks

“…ML-based reconstruction approaches using GNNs [19][20][21][22][23] have been proposed for various tasks in particle physics [24], including tracking [25][26][27][28][29], jet finding [30][31][32] and tagging [33][34][35][36], calorimeter reconstruction [37], pileup mitigation [38], and PF reconstruction [39][40][41]. The clustering of energy deposits in detectors with a realistic, irregulargeometry detector using GNNs has been first proposed in Ref.…”