The first task of this algorithm is to initialize the activation of
all units. This is necessary each time a new pattern is loaded to the
network. The ART2 net is initialized for a new pattern now. The output
and activation will be updated with synchronous propagations until a
stable state is reached. One synchronous propagation cycle means that
each neuron calculate its output and then its new activation. The
required parameters are 
, a, b, c, 
in field1,
field2, field3, field4, field5 respectively.