Path: utzoo!attcan!uunet!ns-mx!iowasp.physics.uiowa.edu!maverick.ksu.ksu.edu!zaphod.mps.ohio-state.edu!ncar!boulder!ccncsu!longs.LANCE.ColoState.EDU!rl980421
From: rl980421@longs.LANCE.ColoState.EDU (Ren-jean Liou)
Newsgroups: comp.ai.neural-nets
Subject: Need help on Hopfield Nets.
Message-ID: <9743@ccncsu.ColoState.EDU>
Date: 27 Sep 90 23:13:51 GMT
Sender: news@ccncsu.ColoState.EDU
Reply-To: rl980421@longs.LANCE.ColoState.EDU (Ren-jean Liou)
Organization: Engineering College, Colorado State University
Lines: 23


      I am doing a project which uses Hopfield network to do pattern
classification.
I wrote a program for the binary case of Hopfield nets. It does not work
very well
especially the capacity is poor. I followed Lippman's paper and did the
exactly same
experiment, i.e., the recognition of a noisy '3', but I could not get
the same result.
I also followed Psaltis et.al. example described in "Optical
implementation of Hopfield
Network" in "Applied Optics" 1985. Still I could not get the same result
and I felt that
their illustration does not make sense.
      I have looked at many of the available NN packages through FTP but
they are
very big in size and is not suitable for me. Is there anyone ever
developed a Hopfield
Network which works OK?(sure) I am wondering if anyone can give me some
advices or even
generously give me the program. I will describe what I have done in more
detail to
you. Thanks a lot.