The Large High Altitude Air Shower Observatory (LHAASO), is a multi-component experiment located at Daocheng (4410 m a.s.l.), Sichuan province, P.R. China. The identification of gamma rays from protons is an important foundation and premise for gamma ray research. In this paper, we use deep learning algorithm to extract the key features of events directly based on a large amount of original information, and explore the identification power of gamma rays from protons of LHAASO experiment. The Convolutional Neural Network(CNN), Deep Neural Networks(DNN) and Graph Neural Networks (GNN) are trained and tested based on a large number of simulation events respectively. Compared with the traditional methods, we have found that the trained CNN, DNN and GNN models all have improvements in the effect of proton and gamma discrimination.
|Subcategory||Experimental Methods & Instrumentation|