Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!swrinde!elroy.jpl.nasa.gov!decwrl!mcnc!gatech!usenet.ins.cwru.edu!ysub!psuvm!barilvm!bimacs!guedalia
From: guedalia@bimacs.BITNET (David Guedalia)
Newsgroups: comp.ai.neural-nets
Subject: Re: Extracting Rules from a Trained Network
Message-ID: <3189@bimacs.BITNET>
Date: 21 Apr 91 14:24:51 GMT
References: <886@uqcspe.cs.uq.oz.au>
Reply-To: guedalia@bimacs.UUCP (David Guedalia)
Organization: Bar-Ilan University, Israel.
Lines: 23

In article <886@uqcspe.cs.uq.oz.au> bakker@cs.uq.oz.au writes:
>I am interested in pointers to any articles, researchers, papers,
>or books that have investigated the extraction of rules from a
>successfully trained neural network.
>
>Does anyone know if this is indeed possible? Can human-readable
>rules be deduced from the distributed weights and connections of a
>neural network?
>
    What type of network are you talking about. I remember some
mention on this board about using a neural net. in a expert system,
would that be the same?
      In a Kohonen feature map the weights would not
say much.  But the distribution of the weights in the map
should have some meaning. Has anyone heard or have
any ideas about how one could represent a feature map not by its
weights but by the relationship between its neighborhoods ?
     I have seen something called instars and out-stars an out-star
could be a feature map and an instar would be the oppositte a way of
representing the feature map by a single vector.  Has anyone seen
refrences to that?


     david