Impact parameter determination using Machine learning algorithms

Determination of impact parameter is a challenge for experimentalists because it cannot be measured directly. Using transport model simulations, it has been demonstrated that machine learning algorithms can infer impact parameter from experimental observables. However, unlike simulations, detection limitations such as geometric efficiencies and energy thresholds of the detector may affect the ability of machine learning to infer impact parameters. In this talk, we will discuss the impact parameter determination using machine learning in the collisions of $^{\mathrm{132}}$Sn$+^{\mathrm{124}}$Sn system at 270 MeV/u using charged particles detected by the S$\pi $RIT Time Projection Chamber. We will also discuss how we validate that the impact parameters determined from machine learning is indeed more accurate than that obtained by various experimental observables.