Reconstructing particles in jets using set transformer and hypergraph prediction networks

Abstract

The task of reconstructing particles from low-level detector response data to predict the set of final state particles in collision events represents a set-to-set prediction task requiring the use of multiple features and their correlations in the input data. We deploy three separate set-to-set neural network architectures to reconstruct particles in events containing a single jet in a fully-simulated calorimeter. Performance is evaluated in terms of particle reconstruction quality, properties regression, and jet-level metrics. The results demonstrate that such a high dimensional end-to-end approach succeeds in surpassing basic parametric approaches in disentangling individual neutral particles inside of jets and optimizing the use of complementary detector information. In particular, the performance comparison favors a novel architecture based on learning hypergraph structure, HGPflow, which benefits from a physically-interpretable approach to particle reconstruction.

17 pages, 21 figures

References in corpus (6)

• Particle-flow reconstruction and global event description with the CMS detector
• Graph Neural Networks in Particle Physics
• MLPF: Efficient machine-learned particle-flow reconstruction using graph neural networks
• Object condensation: one-stage grid-free multi-object reconstruction in physics detectors, graph and image data
• Towards a Computer Vision Particle Flow
• Machine Learning for Particle Flow Reconstruction at CMS

Cited by in corpus (13)

• Improved particle-flow event reconstruction with scalable neural networks for current and future particle detectors
• CaloGraph: Graph-based diffusion model for fast shower generation in calorimeters with irregular geometry
• Portable acceleration of CMS computing workflows with coprocessors as a service
• Set-Conditional Set Generation for Particle Physics
• Advancing Set-Conditional Set Generation: Diffusion Models for Fast Simulation of Reconstructed Particles
• Interplay of Traditional Methods and Machine Learning Algorithms for Tagging Boosted Objects
• One-to-one correspondence reconstruction at the electron-positron Higgs factory
• Progress in End-to-End Optimization of Detectors for Fundamental Physics with Differentiable Programming
• Parnassus: An Automated Approach to Accurate, Precise, and Fast Detector Simulation and Reconstruction
• Fine-tuning machine-learned particle-flow reconstruction for new detector geometries in future colliders
• Reconstructing short-lived particles using hypergraph representation learning
• PASCL: Supervised Contrastive Learning with Perturbative Augmentation for Particle Decay Reconstruction
• TIGER: A Topology-Agnostic, Hierarchical Graph Network for Event Reconstruction