Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!csd4.milw.wisc.edu!bionet!apple!bloom-beacon!bu-cs!bu-it!merrill From: merrill@bucasb.bu.edu (John Merrill) Newsgroups: comp.ai.neural-nets Subject: Re: 3-Layer versus Multi-Layer Message-ID: Date: 27 Jun 89 18:56:01 GMT References: <3417@cosmo.UUCP> Organization: Boston University Center for Adaptive Systems Lines: 19 In-reply-to: jochenru@cosmo.UUCP's message of 20 Jun 89 01:19:09 GMT One reference to such a result is Funahashi, K. (1989). "On the Approximate Realization of Continuous Mappings by Neural Networks", {\bf Neural Networks} (2) 183-192. There are actually several different theorems which prove the same thing, but Funahashi's is the first that I know of which does it from standard sigmoid semi-linear nodes. -- John Merrill | ARPA: merrill@bucasb.bu.edu Center for Adaptive Systems | 111 Cummington Street | Boston, Mass. 02215 | Phone: (617) 353-5765 -- John Merrill | ARPA: merrill@bucasb.bu.edu Center for Adaptive Systems | 111 Cummington Street | Boston, Mass. 02215 | Phone: (617) 353-5765