Track reconstruction performance optimization using machine learning for the upgrade of the ALICE experiment

With the preparation of the future ALICE3 detector planned for the LHC Run 5 and 6, the use of new reconstruction algorithms, more adapted to its new geometry and more efficient, is needed. This project is about optimizing the performances for the charged particle track reconstruction in high-multiplicity Pb-Pb collision environment, in particular, for the particles with low transverse momenta, utilizing the machine learning based Optuna optimization framework to auto-tune the input parameters of the Combinatorial Kalman Filter algorithm within the ACTS track reconstruction software.