Neural networks technique based signal - from - background separa - tion and design optimization for a W / quartz fiber calorimeter

X iv :h ep -e x/ 03 03 02 1v 1 1 4 M ar 2 00 3 UA-NPPS/03/2003 hep-ex/0303021 Neural networks technique based signal-from-background separation and design optimization for a W/quartz fiber calorimeter G. Mavromanolakis ∗ University of Athens, Physi s Department Nu lear and Parti le Physi s Division Panepistimioupoli Ilisia, 15771, Athens, Gree e We present a signal-from-ba kground separation study based on neural networks te hnique applied to a W/quartz ber alorimeter. The performan e in terms of signal e ien y and improvement of the signalto-ba kground ratio are presented. We on lude that by using neural networks we an e iently separate signal from ba kground and a hieve a signal enhan ement over the ba kground of the order of several thousands at high e ien y.