Real-Time SiPM Pulse Deconvolution for a High-Granularity Dual-readout Calorimeter with Neural Networks

This paper presents a neural network approach for real-time deconvolution of Silicon Photomultiplier (SiPM) pulses in the High-Granularity Dual-Readout Calorimeter (HG-DREAM) designed for the Future Circular Collider (FCC-ee). The calorimeter achieves high energy resolution through simultaneous measurement of Cherenkov and scintillation signals combined with longitudinal segmentation via timing measurements. Machine learning techniques are applied to distinguish multiple energy deposits along optical fibers based on signal arrival times, enabling fine three-dimensional shower imaging for particle physics experiments.