home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.ai.neural-nets
- Path: sparky!uunet!cs.utexas.edu!torn!csd.unb.ca!news.ucs.mun.ca!pooh.engr.mun.ca!tapas
- From: tapas@pooh.engr.mun.ca (Tapas Banerjee)
- Subject: HELP - Continous Hopfield Connection
- Message-ID: <tapas.725949309@femto.engr.mun.ca>
- Summary: Problem with connection wts in hopfield networks.
- Keywords: Hopfield, Connection Matrix
- Sender: usenet@news.ucs.mun.ca (NNTP server account)
- Organization: Memorial University of Newfoundland
- Date: Sat, 2 Jan 1993 04:35:09 GMT
- Lines: 26
-
- Dear Netters,
-
- Happy new year to all !!
-
- I am having some problem with Hopfield auto-associative networks.
-
- When I look into binary hopfield with inputs 1,0 - I find that to implement
- that, the connection between neurons has to switch from excitatory to
- inhibitory depending on the output. Precisely, the interaction from Neuron j
- to Neuron i is Tij * (2Xj-1) i.e. with a Tij of say 1, the interaction is
- excitatory with Neuron j output Xij of 1 - but it becomes inhibitory when
- Neuron j output is 0.
-
- Does this remain same with continous hopfield ? i.e. does one have to switch
- between excitation and inhibition depending on neuron outputs ??
-
- Is there some way out with which one can implement binary hopfield with 1,0
- i/p's so that one doesn't have to switch between excitation and inhibition ??
-
- Any help or pointers will be greatly appreciated.
-
- Thanking in advance,
-
- - TAPAS BANERJEE
-
- (tapas@pooh.engr.mun.ca)
-