Gaussian Process Regression as a Sustainable Data-driven Background Estimate Method at the (HL)-LHC

2025/03/10 by Barr, Jackson, Liu, Bingxuan · 3 citations
#FOS: Physical sciences #High Energy Physics - Experiment (hep-ex) #High Energy Physics - Phenomenology (hep-ph)

paper · doi:10.48550/arxiv.2503.07289

Abstract

In this article, we evaluate the performance of a data-driven background estimate method based on Gaussian Process Regression (GPR). A realistic background spectrum from a search conducted by CMS is considered, where a large sub-region below the trigger threshold is included. It is found that the L₂ regularisation can serve as a set of hyperparameters and control the overall modelling performance to satisfy common standards established by experiments at the Large Hadron Collider (LHC). In addition, we show the robustness of this method against increasing luminosity via pseudo-experiments matching the expected luminosity at the High-Luminosity LHC (HL-LHC). While traditional methods relying on empirical functions have been challenged during LHC Run 2 already, a GPR-based technique can offer a solution that is valid through the entire lifetime of the (HL)-LHC.

Cited by

• Enhancing di-jet resonance searches via a final-state radiation jet tagging algorithm
• Log Gaussian Cox Process Background Modeling in High Energy Physics
• Data-Driven High-Dimensional Statistical Inference with Generative Models

Related

• Machine learning method for enforcing variable independence in background estimation with LHC data: ABCDisCoTEC
• Using holistic event information in the trigger
• Modeling Smooth Backgrounds and Generic Localized Signals with Gaussian Processes
• Review of Machine Learning for Real-Time Analysis at the Large Hadron Collider experiments ALICE, ATLAS, CMS and LHCb
• 21-cm Signal from the Epoch of Reionization: A Machine Learning upgrade to Foreground Removal with Gaussian Process Regression