Enhancing Hadron Reconstruction in the CMS High Granularity Calorimeter: A Semi-Parametric Deep Learning Approach for the HL-LHC Era

Yu-Wei Kao^1*, Pedro Manuel Vieira de Castro Ferreira da Silva², Kai-Feng Chen¹

¹Physics, National Taiwan University, Taipei, Taiwan
²CERN, Geneva, Switzerland

* Presenter:Yu-Wei Kao, email:d07222014@ntu.edu.tw

The high-luminosity Large Hadron Collider (HL-LHC), expected to begin operations in 2029, presents significant challenges for particle detection and reconstruction, particularly in terms of processing speed and precision. To address these challenges in hadron reconstruction for the CMS high granularity calorimeter (HGCAL), an innovative silicon-based sampling calorimeter, we employ a semi-parametric regression technique to optimize the hadron energy scale. The regression aims to optimize a linear-like correction using a deep neural network (DNN), providing accurate energy predictions. This algorithm will be integrated into The Iterative Clustering (TICL) framework, enabling comprehensive particle reconstruction in the novel calorimeter.

Keywords: Compact Muon Solenoid, High Granularity Calorimeter, Hadron Energy Regression, Machine Learning