Determination of the impact parameter in high-energy heavy-ion collisions via deep learning *

Abstract In this study, Au+Au collisions with an impact parameter of fm at \sqrt{s_{NN}} = 200 GeV are simulated using the AMPT model to provide preliminary final-state information. After transforming information into appropriate input data (the energy spectra charged hadrons), we construct a multi-layer perceptron (MLP) and convolutional neural network (CNN) connect observables parameters. The results show that both MLP CNN can reconstruct parameters mean absolute error approximately 0.4 fm, although behaves slightly better. Subsequently, test networks different beam energies pseudorapidity ranges in task. These two models work well low high energies. However, when conducting for larger window, exhibits higher prediction accuracy than MLP. Using Grad-CAM method, shed light on 'attention' mechanism model.