A neural discriminator capable to identify impurities in the data sample

Neural networks are applied to a particle discrimination problem in high-energy physics. Information from a specific detector that measures the energy of the incoming particles (a calorimeter) is used to feed the input nodes of the discriminator for the identification of electrons, pions and muons. During the training phase, the neural discriminator was capable to identify impurities in the original data sample obtained from particle beams and this capability was cross checked with a classical method. Having such impurities removed, the discriminator achieved efficiencies of 99.6% (pions), 99.5% (muons) and 98.3% (electrons). The system may be implemented in fast digital signal processor technology envisaging online operation