DeepCore: Convolutional Neural Network for high~$p_T$ jet tracking

A Convolutional Neural Network-based algorithm, DeepCore, has been developed to improve the tracking performance in the jet cores regions. The DeepCore algorithm performs the seeding for the Combinatorial Kalman Filter of the CMS reconstruction in high energy jet cores using the informations of the four pixel detector layers. The preliminary performance of the DeepCore algorithm are presented integrated in the CMS reconstruction.