The problem of extracting a useful signal (a response) buried in relat
ively high amplitude noise has been investigated, under the conditions
of low signal-to-noise ratio. In particular, we present a method for
detecting the ''true'' response of the brain resulting from repeated a
uditory stimulation, based on selective averaging of single-trial evok
ed potentials. Selective averaging is accomplished in two steps. First
, an unsupervised fuzzy-clustering algorithm is employed to identify g
roups of trials with similar characteristics, using a performance inde
x as an optimization criterion. Then, typical responses are obtained b
y ensemble averaging of all trials in the same group. Similarity among
the resulting estimates is quantified through a synchronization measu
re, which accounts for the percentage of time that the estimates are i
n phase, The performance of the classifier is evaluated with synthetic
signals of known characteristics, and its usefulness is demonstrated
with real electrophysiological data obtained from normal volunteers.