Multi-Modal Graph Neural Network Tracking Reconstruction at the Belle II Experiment

In this thesis, I present an end-to-end multi-modal reconstruction algorithm for the Central Drift Chamber and Silicon Vertex Detector of the Belle II experiment at the SuperKEKB collider, using Graph Neural Networks to reconstruct an unknown number of particles.
The algorithm uses hits from both detectors as inputs without pre-filtering to simultaneously predict the number of track candidates in an event and their kinematic properties. In a second step, detector hits are clustered for each track candidate to pass to a track fitting algorithm.
Using a realistic full detector simulation including beam-induced backgrounds and detector noise taken from actual collision data, I find significant improvements in track purity for tracks in a variety of different event topologies compared to the existing baseline algorithm used in Belle II. This is the first end-to-end multi-modal machine learning algorithm for a drift chamber and silicon-based detector that has been utilized in a realistic particle physics environment.