home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.ai.neural-nets
- Path: sparky!uunet!wupost!zaphod.mps.ohio-state.edu!caen!aumeier
- From: aumeier@engin.umich.edu (Steven E Aumeier)
- Subject: ANN State of the Art ?
- Message-ID: <D6r-8k+@engin.umich.edu>
- Date: Fri, 11 Sep 92 21:52:40 EDT
- Organization: University of Michigan Engineering, Ann Arbor
- Nntp-Posting-Host: syndicoot.engin.umich.edu
- Lines: 21
-
- I have two general questions reqarding neural net
- state of the art:
- 1) Does there exist a method by which we are
- able to systematically determine network
- topology (# of nodes and # of layers) for
- a feed-foreword perceptron-type net (not
- a net used for pattern classification where
- a decision tree is mapped onto a neural net).
-
- 2) Does there exist a neural net structure that
- accounts for a priori knowledge of system equations
- (i.e. we may have some knowledge of how an
- input vector is related to a corresponding output
- vector).
-
- If anybody has an answer to these questions, I would certainly
- appreciate hearing from you via E-Mail, the address is given below.
-
- SE Aumeier
- aumeier@syndicoot.engin.umich.edu
-
-