Speaker
Description
Quarkonium production is one of the essential probes for studying the properties of the quark-gluon plasma (QGP) created in relativistic heavy-ion collisions. The suppression of J$/\psi$ meson due to colour screening in medium and medium-induced dissociation was initially proposed as direct evidence of QGP formation. The non-prompt component of J$/\psi$ production from b-hadron decays allows one to access the interaction of beauty quarks with the QGP. The main challenge to extend the measurements to low $p_{\rm T}$ is the huge combinatorial background. The machine learning approach has been used in this data analysis to improve J/$\psi$ meson signal-to-background ratio.
In this poster, the performance of the new machine learning approach, applied on data of Pb--Pb collisions at $\sqrt{s_{\rm NN}}$ = 5.02 TeV, will be presented. The signal reconstruction efficiency and signal over background ratio, obtained from machine learning approach, will be compared with the classical cut approach, as a function of centrality and $p_{\rm T}$. A similar comparison between the two methods will be eventually discussed for efficiency corrected quantities, such as yields.
