A method for adaptively learning is described that effectively and efficiently
uses large amounts of data. Specifically, in one example, a scheme is described
for determining whether to use data to adaptively learn parameters, or whether
to discard the data. In this way, convergent data learning is possible.