Boosted Higgs boson jet reconstruction via a graph neural network

By representing each collider event as a point cloud, we adopt the Graphic Convolutional Network (GCN) with focal loss to reconstruct the Higgs jet in it. This method provides higher Higgs tagging efficiency and better reconstruction accuracy than the traditional jet substructure method. We find the GCN, which is trained on the events of the $H+$jets process, is also applicable to the Higgs jet reconstruction in events of other processes, as long as there are no boosted particles other than the Higgs. Moreover, the features learned by the GCN are complementary to the traditional jet substructure variables in signal and background discrimination.