Loading ...
Sorry, an error occurred while loading the content.

Re: Reply to older topic, self org. NN

Expand Messages
  • g_ai
    wizard_of_frozzbozz wrote: > However, I would find it more interesting to know just > what rules you apply to creating more neurons. Number
    Message 1 of 1398 , Nov 24, 1999
      wizard_of_frozzbozz wrote:<br>> However, I
      would find it more interesting to know just<br>>
      what rules you apply to creating more
      neurons.<br><br>Number of neurons is growing up until the external
      criterion value is decrease. The Group Method of Data
      Handling (GMDH) is based on sorting-out of gradually
      complicated models and evaluation of them by external
      criterion, which is calculated on separate part of data
      sample. The effective input variables, number of layers
      and neurons in hidden layers, optimal model structure
      are determined automatically. Automatic selection is
      based on that fact that external criterion
      characteristic have minimum during complication of model
      structure.<br>In such way computer can found optimal solutions
      itself using minimal apriori information.<br><br>> If
      this<br>> question is answered sufficiently withing one of
      the articles on your<br>> site, I would greatly
      appreciate it if you said which one.<br><br>Please check out
      GMDH survey at
      <br><a href=http://skyscraper.fortunecity.com/cray/793/articles/_review.zip target=new>http://skyscraper.fortunecity.com/cray/793/articles/_review.zip</a><br><br>> Also, in your NN, is there a central area where
      all the<br>> other sections of the NN can
      "communicate",<br><br>Recent developments of the GMDH have led to neuronets
      with active neurons, which realize twice-multilayered
      structure: neurons are multilayered and they are connected
      into multilayered structure. This gives possibility to
      optimize the set of input variables at each layer, while
      the accuracy increases. Not only GMDH algorithms, but
      many modeling or pattern recognition algorithms can be
      used as active neurons. Its accuracy can be increased
      in two ways:<br> - each output of algorithm (active
      neuron) generate new variable, which can be used as a new
      factor in next layers of neuronet;<br> - the set of
      input factors can be optimized at each layer. In usual
      once-multilayered NN the set of input variables can be chosen once
      only. The output variables of previous layers in such
      networks are very effective secondary inputs for the
      neurons of next layers.<br><br> Neuronets with active
      neurons and basic GMDH algorithms were described in<br>
      Selforganization of Neuronets with Active Neurons.<br>
      Ivakhnenko,A.G.,Ivakhnenko,G.A.,Muller,J.A. Pattern recognition and image analysis, 1994,
      vol.4, no.2;<br> Self-Organization of Nets of Active
      Neurons.<br> Ivakhnenko A.G., Muller J.-A. SAMS, 1995, vol.20,
      pp.93-106.<br><br> The self-organizing GMDH approach is described at
      <a href=http://www.stormloader.com/gmdh/index.html target=new>http://www.stormloader.com/gmdh/index.html</a><br><br> Regards,<br><br> Gregory
      Ivakhnenko<br><br><br><br>--<br>Nat.Inst.for Strategic Studies<br>Kyiv, Ukraina
    • smartxpark
      Hello, Have you succeeded ... God To Create Silicon
      Message 1398 of 1398 , Aug 19, 2005

        Have you succeeded

        --- In artificialintelligencegroup@yahoogroups.com, Spider_Plant9 wrote:
        > The rosetta stone is a archeoliguist's dream come
        > true. The Stone was found in a place called rosetta
        > from what I remember but it helped solve the
        > hieroglyphics puzzle because on the stone was written the same
        > text in three languages Greek, Egytian, and Latin. We
        > knew both Greek and Latin. The egytian language though
        > unknown was also the same texts as the known languages.
        > So, using the other two languages we were able to
        > piece together egyptian.<br><br>Now, you can guess what
        > I'm going to do with the texts and languages I
        > collect. I'm going to create a computerized version of the
        > rosetta complete with the "thought" processes to use it
        > directly and extrapolate more words and
        > sentenses.<br><br>Spider_plant9<br>Archeology and the "Words of
        God"<br>To Create Silicon
        > Life<br>Anybody know of a set of standardized interpretations for
        > the Qur'an.
      Your message has been successfully submitted and would be delivered to recipients shortly.