Path: utzoo!attcan!uunet!tut.cis.ohio-state.edu!osu-20.ircc.ohio-state.edu!whiteley-j
From: WHITELEY-J@osu-20.ircc.ohio-state.edu (J. Whiteley)
Newsgroups: comp.ai.neural-nets
Subject: Reference for sufficiency of two layers
Message-ID: <12626564934017@osu-20.ircc.ohio-state.edu>
Date: 2 Oct 90 12:32:44 GMT
Sender: news@tut.cis.ohio-state.edu
Lines: 16

I'm looking for a reference which shows that two layers are sufficient
to approximate any function using a BP network.  All I've dug up so
far is:

Cybenko, G.  Approximation by superpositions of a sigmoidal function.
Unpublished manuscript, Oct. 1988.

Can someone point me to some other references.  Also, is Cybenko's (or 
any other of the references I'm looking for) proof limited to networks
which use the sigmoid or is it more general similar to Kolmogorov's
proof for three layers?

Many thanks for any help!!!

--Rob Whiteley
  (whiteley-j@osu-20.ircc.ohio-state.edu)