Measurements of Higgs bosons decaying into $b\bar{b}$ or $c\bar{c}$ pairs, and how to improve them with GNN-based flavour tagging techniques.

The results from two analyses of the Higgs boson decaying into pairs of $b\bar{b}$ or $c\bar{c}$ quarks and produced in association with a $W$ or $Z$ boson decaying into leptons are briefly presented. The measurements used proton-proton collisions, corresponding to an integrated luminosity of $139 fb^{-1}$, collected at a centre-of-mass energy of $\sqrt{s}=13$ TeV between 2015 and 2018 by the ATLAS experiment at the Large Hadron Collider. The analyses were conducted in different transverse momentum regimes of the vector gauge bosons, with the decay products of the Higgs bosons reconstructed either as pairs of resolved hadronic jets, or as a single larger radius jet. The two measurements heavily relied on efficient jet flavour-tagging. In addition, both employed a method known as truth jet flavour-tagging to enhance the statistical power of the simulated background samples. New, more sophisticated techniques, based on the use of Graph Neural Networks, are shown to improve the overall performance of jet flavour-tagging algorithms and of the truth-tagging method. They could extend the sensitivity of the ATLAS experiment to the analysed physics processes.