Analysis of the electron/pion separation capability with real size ALICE TRD prototypes using a neural network algorithm

In 2004 a complete stack of ALICE TRD final chambers has been tested at CERN. Here, we report on first results of its capability for separating electrons and pions using a neural network algorithm in comparison to a likelihood method on total charge. Measurements for particle momenta from 2 to 10 GeV / c were carried out. As shown for small prototype chambers during a test beam in 2002, an improvement of the pion rejection was also obtained for the entire stack.