Feed-forward neural networks (FFNN) have been used in many HEP applications with spectacular success; precision measurement of the top quark mass [1], and mass limits from leptoquark searches [2] at
the D0 experiment are two very good examples. We explore here more sophisticated ways of using neural networks with ensembles of FFNN - trees, stacks and committees. We will present results from Monte Carlo
studies of applications to searches of new particles such as the Higgs boson and Techni-rho and, to function approximations.
[1] P.C. Bhat, H.B. Prosper and S. S. Snyder, Int. J. Mod. Phys. A13, 5113 (1998).
[2] D0 Collaboration, Phys. Rev. Letters {80} 2051 (1998).