home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!dtix!darwin.sura.net!mips!zaphod.mps.ohio-state.edu!cs.utexas.edu!qt.cs.utexas.edu!yale.edu!yale!gumby!destroyer!ncar!noao!amethyst!organpipe.uug.arizona.edu!organpipe.uug.arizona.edu!bill
- From: bill@nsma.arizona.edu (Bill Skaggs)
- Newsgroups: comp.ai.neural-nets
- Subject: Re: Neural Nets and Brains
- Message-ID: <BILL.92Jul23135614@ca3.nsma.arizona.edu>
- Date: 23 Jul 92 20:56:14 GMT
- References: <1992Jul21.162033.57397@cc.usu.edu> <1992Jul23.013755.18847@hubcap.clemson.edu>
- <arms.711907358@spedden>
- Sender: news@organpipe.uug.arizona.edu
- Organization: ARL Division of Neural Systems, Memory and Aging, University of
- Arizona
- Lines: 31
- In-Reply-To: arms@cs.UAlberta.CA's message of 23 Jul 92 16: 02:38 GMT
-
- arms@cs.UAlberta.CA (Bill Armstrong) writes:
-
- >First off, isn't it rather strange that the most widespread
- >artificial model of neural operation. the multilayer perceptron,
- >uses continuous quantities on its connections, while the dendrites
- >and axons of neurons use "zero or one" type action potentials?
-
- It is true that multilayer perceptrons use continuous-valued signals,
- but there are many artificial models around that use binary-valued
- signals, e.g. Hopfield nets, Boltzmann machines, Kanerva's Sparse
- Distributed Memory, etc., etc.. As a matter of fact, the
- McCulloch-Pitts model used binary neurons.
-
- >Until physiological psychologists start studying adaptive logic
- >networks, can anyone expect much progress on understanding the
- >brain?
-
- Well, the McCulloch-Pitts model was used extensively, and very
- productively, as a model of the brain, so this isn't too convincing.
-
- Any abstract model includes some features and leaves out others.
- Which features will turn out to be crucial can only be known in
- retrospect. There is no reason to think that continuous models are
- incapable of shedding any light on nervous systems. For theoretical
- work they have some real advantages -- among the most important being
- that they make possible certain learning rules, such as backprop, that
- cannot be used with binary models.
-
- "Let a hundred flowers bloom."
-
- -- Bill
-