Comparing Point Cloud Strategies for Collider Event Classification

Abstract

In this paper, we compare several event classification architectures defined on the point cloud representation of collider events. These approaches, which are based on the frameworks of deep sets and edge convolutions, circumvent many of the difficulties associated with traditional feature engineering. To benchmark our architectures against more traditional event classification strategies, we perform a case study involving Higgs boson decays to tau leptons. We find a 2.5 times increase in performance compared to a baseline ATLAS analysis with engineered features. Our point cloud architectures can be viewed as simplified versions of graph neural networks, where each particle in the event corresponds to a graph node. In our case study, we find the best balance of performance and computational cost for simple pairwise architectures, which are based on learned edge features.

19 pages, 7 figures, 3 tables, code available at https://github.com/DelonShen/classifying-collider-events-with-point-clouds; v2: to match published version

References in corpus (5)

• An Introduction to PYTHIA 8.2
• Search for the $b\bar{b}$ decay of the Standard Model Higgs boson in associated $(W/Z)H$ production with the ATLAS detector
• Evidence for the $H\to b\bar{b}$ decay with the ATLAS detector
• Lorentz Boost Networks: Autonomous Physics-Inspired Feature Engineering
• IRC-safe Graph Autoencoder for unsupervised anomaly detection

Cited by in corpus (4)

• Exploring QCD matter in extreme conditions with Machine Learning
• Hypergraphs in LHC Phenomenology -- The Next Frontier of IRC-Safe Feature Extraction
• A rotation-equivariant graph neural network for learning hadronic SMEFT effects
• Foundations of automatic feature extraction at LHC--point clouds and graphs