Detector optimization based on artificial neural network training

We apply artificial neural networks for event-wise analysis of model data for a microchannel plate detector. Based on this data, we estimate an impact parameter, and the collision point coordinates for each event. We have performed the analysis based on several Monte-Carlo collision models. Even though the quality of the existing models is not sufficient for a reliable, model-independent estimation of the collision parameters, the proposed method of parameter reconstruction allows us to estimate the optimal technical characteristics of the detector.