Jet flavor tagging with Particle Transformer for Higgs factories

Abstract

We study the performance of the Particle Transformer (ParT) for jet flavor tagging using ILD full simulation events (1M jets) as well as fast simulation samples (10M and 1M jets). We perform 3-category ($b/c/d$), 6-category ($b/c/d/u/s/g$), and 11-category trainings (including quark--antiquark separation), incorporating multivariate hadron particle identification information from $dE/dx$ and time-of-flight. For $b$/$c$ tagging, we observe a factor of 5--10 improvement over previous BDT-based taggers, and we obtain reasonable performance for strange tagging and quark/antiquark separation. The 10M-jet fast simulation study indicates that further gains are possible with higher training statistics.

Figures (2)

• Figure 1: Performance plots for $b$- and $c$-tagging (three panels shown side-by-side). The line colors indicate background jet categories. The solid, dotted, and dashed curves correspond to ILD full simulation, SGV fast simulation trained with 1M jets, and SGV fast simulation trained with 10M jets, respectively.
• Figure 2: (a) Signal-versus-background performance for 6-category strange tagging in ILD full simulation. (b) Confusion matrix for the 11-category classifier in ILD full simulation, where the predicted flavor is defined as the class with the highest predicted probability.