The neural-network-based second-level trigger developed for the Chooz experiment

The Chooz experiment has recently started data taking and will soon start measuring the neutrino flux from two nuclear power reactors at an average distance of 1 km, aiming at discovering possible neutrino oscillations. A hardware implementation of a neural network algorithm was studied and built in order to rapidly (≈ 200 μs) reconstruct the energy and the position of the events inside the detector. The performances of this system and preliminary results on the first experimental data are presented. The system can be used for a fast rejection of background events.