A growing self-organizing map using a gravitational algorithm was implement
ed on a transputer to study possible separation of GEANT simulated beam bac
kground events and physics events (e(+)e(-) --> B-o(B) over bar(o), q (q) o
ver bar, ggg, 2 gamma,tau(+)tau(-)). A fraction of 75.0% of beam background
events can be rejected, 96.9% of physics events pass the classification. T
he decision time is tau less than or equal to 2 ms, thus the system could b
e used online as level 3 trigger as well as for offline data filtering purp
oses. (C) 1999 Published by Elsevier Science B.V. All rights reserved.