Development and Validation of ML-based trigger algorithms

Complex machine learning (ML) based algorithms are finding their ways to low-level hardware devices like FPGAs. A neural network model is being implemented for the online identification and trigger of the passage of charged particles through a drift-tubes detector and for the reconstruction of its trajectory; such model is being tested on a set of such detectors operating at Legnaro INFN National Laboratory (LNL) with cosmic muons. The ultimate goal is to deploy this algorithm on the trigger system of the muon spectrometer of CMS at the LHC. The thesis work describes the performances of such algorithms, studying the data collected in Legnaro, and further develop it to extend its acceptance.