Applications of deep learning in jet quenching

Deep learning techniques have broad applications in studying jet quenching phenomena. This paper reviews the works in recent years from several scholars employing di fferent data representations of jet samples and architectures of neural network in terms of reconstructing the momentum of jets in a hot and dense QCD medium, distinguishing the vacuum jets and medium jets, and, in particular, identifying the energy loss of jets as well as distinguishing quark- and gluon-initiated jets in a QCD medium. In the study of jet energy loss prediction, we demonstrate that deep learning techniques can identify the degree of energy loss of high-energy jets traversing hot QCD matter on a jet-by-jet basis and show that our method advances the jet tomographic study of hot QCD matter. In the study of distinguishing quark-jets and gluon-jets in the medium, we find that the classification accuracy gradually decreases as the jets lose energy. Lastly, we discuss the medium modifications of quark and gluon jet substructures in a perspective view.