No subject

[Scott.Fahlman@SEF1.SLISP.CS.CMU.EDU]

    I am interested in constructive  algorithms  for  neural  networks 
    (capable of generating new units and layers in the NN).  Is  there 
    any constructive algorithm for  NN  whose  units  have  continuous 
    (e.g.: sigmoidal) activation function?
    
The only nets that come to mind that fit your description are
Cascade-Correlation (Fahlman and Lebiere, paper in NIPS-89 proceedings) and
SONN (Tenrio and Lee, NIPS-88).  There are a lot of system around that
build up a *single* layer of hidden units with continuous activation
functions.

Information on how to FTP the Cascade-Correlation tech report and
simulation code has been sent to this list before.  I'll send it to you in
a separate message.

-- Scott