Abstract
The increasing data rates in modern high-energy physics experiments such as
ALICE at the LHC and the upcoming ePIC experiment at the Electron-Ion Collider
(EIC) present significant challenges in real-time event selection and data
storage. This paper explores the novel application of machine learning
techniques, to enhance the identification of rare low-multiplicity events, such
as ultraperipheral collisions (UPCs) and central exclusive diffractive
processes. We focus on utilising machine learning models to perform early event
classification, even before full event reconstruction, in continuous readout
systems. We estimate data rates and disk space requirements for photoproduction
and central exclusive diffractive processes in both ALICE and ePIC. We show
that machine learning techniques can not only optimize data selection but also
significantly reduce storage requirements in continuous readout environments,
providing a scalable solution for the upcoming era of high-luminosity particle
physics experiments.