Reconstructing the kinematics of deep inelastic scattering with deep learning

We introduce a method to reconstruct the kinematics of neutral-current deep inelastic scattering (DIS) using neural network (DNN). Unlike traditional methods, it exploits full kinematic information both scattered electron and hadronic-final state, accounts for QED radiation by identifying events with radiated photons event-level momentum imbalance. The is studied simulated at HERA future Electron–Ion Collider (EIC). show that DNN outperforms all methods over phase space, improving resolution reducing bias. Our has potential extend reach experiments EIC, thus their discovery in polarized nuclear DIS.