Kink recognition with neural networks

Abstract

The task of finding decays of charged tracks inside a tracking device is divided into two parts. First, a neural network is used to recognize kinks in well-constructed tracks. The inputs to this classification network are the residuals and the curvature obtained by a one-track fit. If a kink has been found, the same inputs are fed into a second neural network, which gives the radial position of the decay vertex. Both algorithms use feedforward nets with error backpropagation. Very good performance is found in comparison with conventional methods.<>