A new, robust and computationally attractive approach to the problem o
f time series classification is discussed in this paper. Both the Baye
sian as well as a new adaptive classification scheme for source select
ion are discussed. Simulation results are included to demonstrate the
effectiveness of the new methodology.