Online neural filtering operating over segmented discriminating components

In high energy collider experiments, the filtering (triggering) systems are responsible for event selection in a huge amount of data generated by particle colliders. In this work a triggering strategy based on segmented principal discriminating components (SPCD) is proposed for the ATLAS detector second-level trigger. A segmented signal processing strategy is proposed here in order to exploit fully the different characteristics of each detector layer. Neural classifiers fed from SPCD perform particle identification. Through the proposed approach, a discrimination efficiency of 97.9% was achieved for a false alarm probability of 2.7%, which outperforms the baseline discriminator in use.