A Trial of Classification of Proton- and Heavy-Ion-Induced Families Using Neural Networks
Abstract
A feed-forward neural network trained using backpropagation is applied to classify proton- and heavy-ion-induced cosmic-ray families. Fifteen input variables which characterize three-dimensional behavior of the families are chosen. The network successfully classify the events with classification efficiency ∼ 85%. The trained neural network is applied for estimating the fraction of heavy-induced events in the simulated families and also in the experimental families observed in the Pamir chambers.
- Publication:
-
Nuclear Physics B Proceedings Supplements
- Pub Date:
- February 1997
- DOI:
- 10.1016/S0920-5632(96)00896-1
- Bibcode:
- 1997NuPhS..52..243T