In the sequel, the problem of selecting the number of units
in a competitive learning net will be addressed in Section
11. Then RPCL is proposed and analyzed in Section 111. In
Section IV, FSCL, and RPCL are experimentally compared
in the problems of unsupervised classificationand RBF net's
training. In Section V, the curve detection problem is modeled
into a problem of competitive learning, and both FSCL and
RPCL are tested with a problem of detecting four lines in a
diamond-shaped frame.