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

Re: [Artificial Intelligence Group] How neural network can achive artificial intelligence?

Expand Messages
  • Pesky Bee
    Connectionists think that intelligence can be achieved by neural networks not because what happens with an individual neuron, but because what emerges from the
    Message 1 of 15 , Jan 3, 2006
    • 0 Attachment
      Connectionists think that intelligence can be achieved by
      neural networks not because what happens with an individual
      neuron, but because what emerges from the operation of
      a large number of them. Even if a the information processing
      of a single neuron hardly can be considered "intelligent",
      the same might not be true about a group of neurons. Of
      course, artificial intelligence using artificial neural
      networks is still a hypothesis, although results so far
      have been promising.

      *PB*



      ----- Original Message -----
      From: "Sathish kumar" <psathishdl@...>
      To: <artificialintelligencegroup@yahoogroups.com>
      Sent: Monday, January 02, 2006 2:07 AM
      Subject: [Artificial Intelligence Group] How neural network can achive
      artificial intelligence?


      > How neural network can achive artificial intelligence?
      > jest by connecting neuron do you think you can achive intelligence.
      >
      > do neuron processing the information ?
      >
      >
      > Send instant messages to your online friends http://in.messenger.yahoo.com
      >
      > [Non-text portions of this message have been removed]
      >
      >
      >
      >
      >
      >
      > Yahoo! Groups Links
      >
      >
      >
      >
      >
      >
    • Dave Faulkner
      Here it is, in a nutshell, as I understand it -- Neural networks , like other mathematical constructions such as fuzzy logic networks, radial basis functions,
      Message 2 of 15 , Jan 3, 2006
      • 0 Attachment
        Here it is, in a nutshell, as I understand it --

        Neural networks , like other mathematical constructions such as
        fuzzy logic networks, radial basis functions, Fourier series, etc.
        belong to a class of functions that are Universal Approximation Functions.

        UFAs are functions like NN that can mimic -- to some degree of
        approximation -- many other well behaved functions, and some
        not so well behaved functions.

        So, an NN can approximate a polynomial, or a sinusoid, or a linear
        switching function if you tweak the parameters properly. The hard
        part is knowing what these tweaking parameter values are, because
        they dictate whether the NN looks like a polynomial or a circle
        or whatever.

        And if you don't know the function, that's ok too because you can
        make the NN resemble any set of XY graphable data -- to some
        degree of resemblance -- simply be having the actual data. The
        Data can be measured or derived or whatever; the NN can be
        made to "fit" the data.

        Key in understanding how this can possibly be "intelligence" is realizing
        that the NN can also mimic very complicated logic.

        For example, Do I need to go grocery shopping?

        I need to go grocery shopping if --

        - there's no food in the house.
        - I can pay for it
        - the car works
        - etc....

        So you can set up the NN to evaluate the situation and give you
        and "intelligent" response --

        NN (Grocery shop) == (No food) AND (Money) AND (Working Car)

        Now imagine 100,000 of such NNs with input to each other and the
        ability to change each others logic parameters and you'll begin to
        imagine a small working brain (like an insects). Most usable and trainable
        NNs are small enough to tackle just one problem, so they resemble
        large equations. But the functions that they can mimic can be
        large and complex, and so appear "intelligent" in the very
        specialized context that they were created in.







        ----- Original Message -----
        From: "Pesky Bee" <peskybee2@...>
        To: <artificialintelligencegroup@yahoogroups.com>
        Sent: Tuesday, January 03, 2006 6:28 AM
        Subject: Re: [Artificial Intelligence Group] How neural network can achive artificial intelligence?


        > Connectionists think that intelligence can be achieved by
        > neural networks not because what happens with an individual
        > neuron, but because what emerges from the operation of
        > a large number of them. Even if a the information processing
        > of a single neuron hardly can be considered "intelligent",
        > the same might not be true about a group of neurons. Of
        > course, artificial intelligence using artificial neural
        > networks is still a hypothesis, although results so far
        > have been promising.
        >
        > *PB*
        >
        >
        >
        > ----- Original Message -----
        > From: "Sathish kumar" <psathishdl@...>
        > To: <artificialintelligencegroup@yahoogroups.com>
        > Sent: Monday, January 02, 2006 2:07 AM
        > Subject: [Artificial Intelligence Group] How neural network can achive
        > artificial intelligence?
        >
        >
        >> How neural network can achive artificial intelligence?
        >> jest by connecting neuron do you think you can achive intelligence.
        >>
        >> do neuron processing the information ?
        >>
        >>
        >> Send instant messages to your online friends http://in.messenger.yahoo.com
        >>
        >> [Non-text portions of this message have been removed]
        >>
        >>
        >>
        >>
        >>
        >>
        >> Yahoo! Groups Links
        >>
        >>
        >>
        >>
        >>
        >>
        >
        >
        >
        >
        >
        >
        > Yahoo! Groups Links
        >
        >
        >
        >
        >
      • Leo Rijadi
        Agree to that. I ve just read Laurene Fausett and Rao & Rao. NN is so fascinating for a beginer like me. After some epoch, NN will converge to some stable
        Message 3 of 15 , Jan 3, 2006
        • 0 Attachment
          Agree to that. I've just read Laurene Fausett
          and Rao & Rao.

          NN is so fascinating for a beginer like me.
          After some epoch, NN will converge to some
          stable weights (the perceptron learning rule
          convergence theorem, Fausett: pp. 76)


          Leo Rijadi

          --- Dave Faulkner <dfaulkne@...> wrote:

          > Here it is, in a nutshell, as I understand it --
          >
          > Neural networks , like other mathematical constructions such as
          > fuzzy logic networks, radial basis functions, Fourier series, etc.
          > belong to a class of functions that are Universal Approximation
          > Functions.
          >
          > UFAs are functions like NN that can mimic -- to some degree of
          > approximation -- many other well behaved functions, and some
          > not so well behaved functions.
          >
          > So, an NN can approximate a polynomial, or a sinusoid, or a linear
          > switching function if you tweak the parameters properly. The hard
          > part is knowing what these tweaking parameter values are, because
          > they dictate whether the NN looks like a polynomial or a circle
          > or whatever.
          >
          > And if you don't know the function, that's ok too because you can
          > make the NN resemble any set of XY graphable data -- to some
          > degree of resemblance -- simply be having the actual data. The
          > Data can be measured or derived or whatever; the NN can be
          > made to "fit" the data.
          >
          > Key in understanding how this can possibly be "intelligence" is
          > realizing
          > that the NN can also mimic very complicated logic.
          >
          > For example, Do I need to go grocery shopping?
          >
          > I need to go grocery shopping if --
          >
          > - there's no food in the house.
          > - I can pay for it
          > - the car works
          > - etc....
          >
          > So you can set up the NN to evaluate the situation and give you
          > and "intelligent" response --
          >
          > NN (Grocery shop) == (No food) AND (Money) AND (Working Car)
          >
          > Now imagine 100,000 of such NNs with input to each other and the
          > ability to change each others logic parameters and you'll begin to
          > imagine a small working brain (like an insects). Most usable and
          > trainable
          > NNs are small enough to tackle just one problem, so they resemble
          > large equations. But the functions that they can mimic can be
          > large and complex, and so appear "intelligent" in the very
          > specialized context that they were created in.
          >
          >
          >
          >
          >
          >
          >
          > ----- Original Message -----
          > From: "Pesky Bee" <peskybee2@...>
          > To: <artificialintelligencegroup@yahoogroups.com>
          > Sent: Tuesday, January 03, 2006 6:28 AM
          > Subject: Re: [Artificial Intelligence Group] How neural network can
          > achive artificial intelligence?
          >
          >
          > > Connectionists think that intelligence can be achieved by
          > > neural networks not because what happens with an individual
          > > neuron, but because what emerges from the operation of
          > > a large number of them. Even if a the information processing
          > > of a single neuron hardly can be considered "intelligent",
          > > the same might not be true about a group of neurons. Of
          > > course, artificial intelligence using artificial neural
          > > networks is still a hypothesis, although results so far
          > > have been promising.
          > >
          > > *PB*
          > >
          > >
          > >
          > > ----- Original Message -----
          > > From: "Sathish kumar" <psathishdl@...>
          > > To: <artificialintelligencegroup@yahoogroups.com>
          > > Sent: Monday, January 02, 2006 2:07 AM
          > > Subject: [Artificial Intelligence Group] How neural network can achive
          >
          > > artificial intelligence?
          > >
          > >
          > >> How neural network can achive artificial intelligence?
          > >> jest by connecting neuron do you think you can achive intelligence.
          > >>
          > >> do neuron processing the information ?
          > >>
          > >>
          > >> Send instant messages to your online friends
          > http://in.messenger.yahoo.com
          > >>
          > >> [Non-text portions of this message have been removed]
          > >>
          > >>
          > >>
          > >>
          > >>
          > >>
          > >> Yahoo! Groups Links
          > >>
          > >>
          > >>
          > >>
          > >>
          > >>
          > >
          > >
          > >
          > >
          > >
          > >
          > > Yahoo! Groups Links
          > >
          > >
          > >
          > >
          > >
          >
          >
          >
          > Yahoo! Groups Links
          >
          >
          >
          >
          >
          >
          >




          __________________________________________
          Yahoo! DSL – Something to write home about.
          Just $16.99/mo. or less.
          dsl.yahoo.com
        • nerosfiddle1
          Wow you guys speak kinda my same language? But I m no programmer. Strange. The vocabulary different, me no math but history. Whole tribes of you/me? My
          Message 4 of 15 , Jan 4, 2006
          • 0 Attachment
            Wow you guys speak kinda my same language? But I'm no programmer.
            Strange. The vocabulary different, me no math but history. Whole
            tribes of you/me? My theories must be altered...thanks. Huh if my
            nano chips were translated we could speak the same language
            ("Tron,"'"I Robot.")

            Hey hire me I'll build you castles in the snow? Whole kingdoms & 200
            computer games at least x 200 million more variables. Much better
            than Nash sorry to say he was my idol? I have only disproved two of
            his non math ones so far. The first major one took my eyes 25
            seconds to reach the meat...Plato disproves Nash.

            Stalin, etc etrc ertc vet cvert5 vcg edfgdfgdftgdfddvdvd




            --- In artificialintelligencegroup@yahoogroups.com, Leo Rijadi
            <leorijadi@y...> wrote:
            >
            > Agree to that. I've just read Laurene Fausett
            > and Rao & Rao.
            >
            > NN is so fascinating for a beginer like me.
            > After some epoch, NN will converge to some
            > stable weights (the perceptron learning rule
            > convergence theorem, Fausett: pp. 76)
            >
            >
            > Leo Rijadi
            >
            > --- Dave Faulkner <dfaulkne@t...> wrote:
            >
            > > Here it is, in a nutshell, as I understand it --
            > >
            > > Neural networks , like other mathematical constructions such as
            > > fuzzy logic networks, radial basis functions, Fourier series,
            etc.
            > > belong to a class of functions that are Universal Approximation
            > > Functions.
            > >
            > > UFAs are functions like NN that can mimic -- to some degree of
            > > approximation -- many other well behaved functions, and some
            > > not so well behaved functions.
            > >
            > > So, an NN can approximate a polynomial, or a sinusoid, or a linear
            > > switching function if you tweak the parameters properly. The hard
            > > part is knowing what these tweaking parameter values are, because
            > > they dictate whether the NN looks like a polynomial or a circle
            > > or whatever.
            > >
            > > And if you don't know the function, that's ok too because you can
            > > make the NN resemble any set of XY graphable data -- to some
            > > degree of resemblance -- simply be having the actual data. The
            > > Data can be measured or derived or whatever; the NN can be
            > > made to "fit" the data.
            > >
            > > Key in understanding how this can possibly be "intelligence" is
            > > realizing
            > > that the NN can also mimic very complicated logic.
            > >
            > > For example, Do I need to go grocery shopping?
            > >
            > > I need to go grocery shopping if --
            > >
            > > - there's no food in the house.
            > > - I can pay for it
            > > - the car works
            > > - etc....
            > >
            > > So you can set up the NN to evaluate the situation and give you
            > > and "intelligent" response --
            > >
            > > NN (Grocery shop) == (No food) AND (Money) AND (Working Car)
            > >
            > > Now imagine 100,000 of such NNs with input to each other and the
            > > ability to change each others logic parameters and you'll begin to
            > > imagine a small working brain (like an insects). Most usable and
            > > trainable
            > > NNs are small enough to tackle just one problem, so they resemble
            > > large equations. But the functions that they can mimic can be
            > > large and complex, and so appear "intelligent" in the very
            > > specialized context that they were created in.
            > >
            > >
            > >
            > >
            > >
            > >
            > >
            > > ----- Original Message -----
            > > From: "Pesky Bee" <peskybee2@g...>
            > > To: <artificialintelligencegroup@yahoogroups.com>
            > > Sent: Tuesday, January 03, 2006 6:28 AM
            > > Subject: Re: [Artificial Intelligence Group] How neural network
            can
            > > achive artificial intelligence?
            > >
            > >
            > > > Connectionists think that intelligence can be achieved by
            > > > neural networks not because what happens with an individual
            > > > neuron, but because what emerges from the operation of
            > > > a large number of them. Even if a the information processing
            > > > of a single neuron hardly can be considered "intelligent",
            > > > the same might not be true about a group of neurons. Of
            > > > course, artificial intelligence using artificial neural
            > > > networks is still a hypothesis, although results so far
            > > > have been promising.
            > > >
            > > > *PB*
            > > >
            > > >
            > > >
            > > > ----- Original Message -----
            > > > From: "Sathish kumar" <psathishdl@y...>
            > > > To: <artificialintelligencegroup@yahoogroups.com>
            > > > Sent: Monday, January 02, 2006 2:07 AM
            > > > Subject: [Artificial Intelligence Group] How neural network can
            achive
            > >
            > > > artificial intelligence?
            > > >
            > > >
            > > >> How neural network can achive artificial intelligence?
            > > >> jest by connecting neuron do you think you can achive
            intelligence.
            > > >>
            > > >> do neuron processing the information ?
            > > >>
            > > >>
            > > >> Send instant messages to your online friends
            > > http://in.messenger.yahoo.com
            > > >>
            > > >> [Non-text portions of this message have been removed]
            > > >>
            > > >>
            > > >>
            > > >>
            > > >>
            > > >>
            > > >> Yahoo! Groups Links
            > > >>
            > > >>
            > > >>
            > > >>
            > > >>
            > > >>
            > > >
            > > >
            > > >
            > > >
            > > >
            > > >
            > > > Yahoo! Groups Links
            > > >
            > > >
            > > >
            > > >
            > > >
            > >
            > >
            > >
            > > Yahoo! Groups Links
            > >
            > >
            > >
            > >
            > >
            > >
            > >
            >
            >
            >
            >
            > __________________________________________
            > Yahoo! DSL – Something to write home about.
            > Just $16.99/mo. or less.
            > dsl.yahoo.com
            >
          • Stan
            For the most part that s true, but only for certian types of networks. Unfortunatly, for recurrent networks or dynamic neuron models, perceptron learning isn t
            Message 5 of 15 , Jan 4, 2006
            • 0 Attachment
              For the most part that's true, but only for certian types of networks.
              Unfortunatly, for recurrent networks or dynamic neuron models,
              perceptron learning isn't much help.

              ~Stan

              --- In artificialintelligencegroup@yahoogroups.com, Leo Rijadi
              <leorijadi@y...> wrote:
              >
              > Agree to that. I've just read Laurene Fausett
              > and Rao & Rao.
              >
              > NN is so fascinating for a beginer like me.
              > After some epoch, NN will converge to some
              > stable weights (the perceptron learning rule
              > convergence theorem, Fausett: pp. 76)
              >
              >
              > Leo Rijadi
              >
              > --- Dave Faulkner <dfaulkne@t...> wrote:
              >
              > > Here it is, in a nutshell, as I understand it --
              > >
              > > Neural networks , like other mathematical constructions such as
              > > fuzzy logic networks, radial basis functions, Fourier series, etc.
              > > belong to a class of functions that are Universal Approximation
              > > Functions.
              > >
              > > UFAs are functions like NN that can mimic -- to some degree of
              > > approximation -- many other well behaved functions, and some
              > > not so well behaved functions.
              > >
              > > So, an NN can approximate a polynomial, or a sinusoid, or a linear
              > > switching function if you tweak the parameters properly. The hard
              > > part is knowing what these tweaking parameter values are, because
              > > they dictate whether the NN looks like a polynomial or a circle
              > > or whatever.
              > >
              > > And if you don't know the function, that's ok too because you can
              > > make the NN resemble any set of XY graphable data -- to some
              > > degree of resemblance -- simply be having the actual data. The
              > > Data can be measured or derived or whatever; the NN can be
              > > made to "fit" the data.
              > >
              > > Key in understanding how this can possibly be "intelligence" is
              > > realizing
              > > that the NN can also mimic very complicated logic.
              > >
              > > For example, Do I need to go grocery shopping?
              > >
              > > I need to go grocery shopping if --
              > >
              > > - there's no food in the house.
              > > - I can pay for it
              > > - the car works
              > > - etc....
              > >
              > > So you can set up the NN to evaluate the situation and give you
              > > and "intelligent" response --
              > >
              > > NN (Grocery shop) == (No food) AND (Money) AND (Working Car)
              > >
              > > Now imagine 100,000 of such NNs with input to each other and the
              > > ability to change each others logic parameters and you'll begin to
              > > imagine a small working brain (like an insects). Most usable and
              > > trainable
              > > NNs are small enough to tackle just one problem, so they resemble
              > > large equations. But the functions that they can mimic can be
              > > large and complex, and so appear "intelligent" in the very
              > > specialized context that they were created in.
              > >
              > >
              > >
              > >
              > >
              > >
              > >
              > > ----- Original Message -----
              > > From: "Pesky Bee" <peskybee2@g...>
              > > To: <artificialintelligencegroup@yahoogroups.com>
              > > Sent: Tuesday, January 03, 2006 6:28 AM
              > > Subject: Re: [Artificial Intelligence Group] How neural network can
              > > achive artificial intelligence?
              > >
              > >
              > > > Connectionists think that intelligence can be achieved by
              > > > neural networks not because what happens with an individual
              > > > neuron, but because what emerges from the operation of
              > > > a large number of them. Even if a the information processing
              > > > of a single neuron hardly can be considered "intelligent",
              > > > the same might not be true about a group of neurons. Of
              > > > course, artificial intelligence using artificial neural
              > > > networks is still a hypothesis, although results so far
              > > > have been promising.
              > > >
              > > > *PB*
              > > >
              > > >
              > > >
              > > > ----- Original Message -----
              > > > From: "Sathish kumar" <psathishdl@y...>
              > > > To: <artificialintelligencegroup@yahoogroups.com>
              > > > Sent: Monday, January 02, 2006 2:07 AM
              > > > Subject: [Artificial Intelligence Group] How neural network can
              achive
              > >
              > > > artificial intelligence?
              > > >
              > > >
              > > >> How neural network can achive artificial intelligence?
              > > >> jest by connecting neuron do you think you can achive
              intelligence.
              > > >>
              > > >> do neuron processing the information ?
              > > >>
              > > >>
              > > >> Send instant messages to your online friends
              > > http://in.messenger.yahoo.com
              > > >>
              > > >> [Non-text portions of this message have been removed]
              > > >>
              > > >>
              > > >>
              > > >>
              > > >>
              > > >>
              > > >> Yahoo! Groups Links
              > > >>
              > > >>
              > > >>
              > > >>
              > > >>
              > > >>
              > > >
              > > >
              > > >
              > > >
              > > >
              > > >
              > > > Yahoo! Groups Links
              > > >
              > > >
              > > >
              > > >
              > > >
              > >
              > >
              > >
              > > Yahoo! Groups Links
              > >
              > >
              > >
              > >
              > >
              > >
              > >
              >
              >
              >
              >
              > __________________________________________
              > Yahoo! DSL – Something to write home about.
              > Just $16.99/mo. or less.
              > dsl.yahoo.com
              >
            • Sharmontime Wonder
              None has commented abt my views. Let alone sayin good, not even one said wrong. If i am incorrect pls correct me. I ll be obliged. Thanking You. Sharmontime.W.
              Message 6 of 15 , Jan 12, 2006
              • 0 Attachment
                None has commented abt my views. Let alone sayin good, not even one
                said wrong.
                If i am incorrect pls correct me.
                I'll be obliged.
                Thanking You.
                Sharmontime.W.




                --- In artificialintelligencegroup@yahoogroups.com, Sharmontime
                Wonder <sharmontime@y...> wrote:
                >
                > the neural networks are to be trained from a, b,c....(for
                whatever the net is to be used for.)
                > then, it is given some situations to act wherein it is suppose
                to display its intelligence.
                > this should be like a child is born, and you teach him the
                basics, and he starts to think. whenever he is wrong, he's either
                corrected by elders or learns that its fire, and i should not touch
                it.
                > this is what i want to simulate.
                >
                >
                > the first sentence i mentioned :....(for whatever the net is to
                be used for.), i meant that presently, i believe neural nets are
                trained for some specific purposes.
                > this is what i said i believe.
                >
                > i have not read much. i have always been fascinated by the terms
                AI, NN and also robotics.
                >
                > i have been thinking various ways of simulating a brain. this
                simulation of the brain and the nervous system is what i understand
                to be the neural networks.
                >
                > training the net is nothing but imbibing (artificial)
                intelligence in it.
                > and embedding this system in electronic circuits and machines
                and etc., is robotics.
                >
                > people interested in further discussions are welcome.
                > please feel free to correct me. i am very much open to all.
                > all sorts of criticisms are welcome.
                >
                > thanks
                >
                > regards
                > - sharmontime
                >
                >
                > Sathish kumar <psathishdl@y...> wrote:
                > How neural network can achive artificial intelligence?
                > jest by connecting neuron do you think you can achive
                intelligence.
                >
                > do neuron processing the information ?
                >
                >
                > Send instant messages to your online friends
                http://in.messenger.yahoo.com
                >
                > [Non-text portions of this message have been removed]
                >
                >
                >
                >
                >
                > SPONSORED LINKS
                > Computer science college Computer science distance
                learning Computer science Artificial intelligence software
                Artificial intelligence in business Artificial intelligence and
                expert system
                >
                > ---------------------------------
                > YAHOO! GROUPS LINKS
                >
                >
                > Visit your group "artificialintelligencegroup" on the web.
                >
                > To unsubscribe from this group, send an email to:
                > artificialintelligencegroup-unsubscribe@yahoogroups.com
                >
                > Your use of Yahoo! Groups is subject to the Yahoo! Terms of
                Service.
                >
                >
                > ---------------------------------
                >
                >
                >
                >
                >
                >
                >
                > ---------------------------------
                > Yahoo! for Good - Make a difference this year.
                >
                > [Non-text portions of this message have been removed]
                >
              • JW
                You can train a NN to do about anything simple , but they are better for Fuzzy Logic type of applications than yes/no thinking. You can just program yes/no
                Message 7 of 15 , Jan 13, 2006
                • 0 Attachment
                  You can train a NN to do about anything "simple", but they are better
                  for Fuzzy Logic type of applications than yes/no thinking. You can
                  just program yes/no thinking, no NN needed.

                  Recent research into the way the human brain works indicates that
                  NN's are not close to human thinking, they just simulate some human
                  thought processes.

                  NN's are already in place and used. Voice recognition (for about $50
                  you can get a really good voice recognition program for you Windows
                  computer now, you just have to train it to your voice), Visual object
                  recognition, I believe grammar checking (correct me on that one if I
                  am wrong), They are working on crash avoidance for cars (if people
                  paid attention to the road when driving, we wouldn't need that one),
                  financial predictions, photo copiers (at least some). Just a few
                  examples.







                  --- In artificialintelligencegroup@yahoogroups.com, "Sharmontime
                  Wonder" <sharmontime@y...> wrote:
                  >
                  > None has commented abt my views. Let alone sayin good, not even one
                  > said wrong.
                  > If i am incorrect pls correct me.
                  > I'll be obliged.
                  > Thanking You.
                  > Sharmontime.W.
                  >
                  >
                  >
                  >
                  > --- In artificialintelligencegroup@yahoogroups.com, Sharmontime
                  > Wonder <sharmontime@y...> wrote:
                  > >
                  > > the neural networks are to be trained from a, b,c....(for
                  > whatever the net is to be used for.)
                  > > then, it is given some situations to act wherein it is suppose
                  > to display its intelligence.
                  > > this should be like a child is born, and you teach him the
                  > basics, and he starts to think. whenever he is wrong, he's either
                  > corrected by elders or learns that its fire, and i should not touch
                  > it.
                  > > this is what i want to simulate.
                  > >
                  > >
                  > > the first sentence i mentioned :....(for whatever the net is to
                  > be used for.), i meant that presently, i believe neural nets are
                  > trained for some specific purposes.
                  > > this is what i said i believe.
                  > >
                  > > i have not read much. i have always been fascinated by the
                  terms
                  > AI, NN and also robotics.
                  > >
                  > > i have been thinking various ways of simulating a brain. this
                  > simulation of the brain and the nervous system is what i understand
                  > to be the neural networks.
                  > >
                  > > training the net is nothing but imbibing (artificial)
                  > intelligence in it.
                  > > and embedding this system in electronic circuits and machines
                  > and etc., is robotics.
                  > >
                  > > people interested in further discussions are welcome.
                  > > please feel free to correct me. i am very much open to all.
                  > > all sorts of criticisms are welcome.
                  > >
                  > > thanks
                  > >
                  > > regards
                  > > - sharmontime
                  > >
                  > >
                  > > Sathish kumar <psathishdl@y...> wrote:
                  > > How neural network can achive artificial intelligence?
                  > > jest by connecting neuron do you think you can achive
                  > intelligence.
                  > >
                  > > do neuron processing the information ?
                  > >
                  > >
                  > > Send instant messages to your online friends
                  > http://in.messenger.yahoo.com
                  > >
                  > > [Non-text portions of this message have been removed]
                  > >
                  > >
                  > >
                  > >
                  > >
                  > > SPONSORED LINKS
                  > > Computer science college Computer science distance
                  > learning Computer science Artificial intelligence software
                  > Artificial intelligence in business Artificial intelligence and
                  > expert system
                  > >
                  > > ---------------------------------
                  > > YAHOO! GROUPS LINKS
                  > >
                  > >
                  > > Visit your group "artificialintelligencegroup" on the web.
                  > >
                  > > To unsubscribe from this group, send an email to:
                  > > artificialintelligencegroup-unsubscribe@yahoogroups.com
                  > >
                  > > Your use of Yahoo! Groups is subject to the Yahoo! Terms of
                  > Service.
                  > >
                  > >
                  > > ---------------------------------
                  > >
                  > >
                  > >
                  > >
                  > >
                  > >
                  > >
                  > > ---------------------------------
                  > > Yahoo! for Good - Make a difference this year.
                  > >
                  > > [Non-text portions of this message have been removed]
                  > >
                  >
                • Narayan
                  HI Friends I am a Java, j2EE developer, I have done my specialisation in AI, but i couldn t find single opportunity to work on AI systems. I have been working
                  Message 8 of 15 , Jan 15, 2006
                  • 0 Attachment
                    HI Friends

                    I am a Java, j2EE developer, I have done my specialisation in AI, but
                    i couldn't find single opportunity to work on AI systems.
                    I have been working on java domain from last 3 years, but still looking for
                    AI opportunity to work.
                    Please help me to get into AI field, if any requirements are there in ur
                    company kindly reffer me.

                    Thanking you
                    Narayan





                    On 1/13/06, JW <johnfr3@...> wrote:
                    >
                    > You can train a NN to do about anything "simple", but they are better
                    > for Fuzzy Logic type of applications than yes/no thinking. You can
                    > just program yes/no thinking, no NN needed.
                    >
                    > Recent research into the way the human brain works indicates that
                    > NN's are not close to human thinking, they just simulate some human
                    > thought processes.
                    >
                    > NN's are already in place and used. Voice recognition (for about $50
                    > you can get a really good voice recognition program for you Windows
                    > computer now, you just have to train it to your voice), Visual object
                    > recognition, I believe grammar checking (correct me on that one if I
                    > am wrong), They are working on crash avoidance for cars (if people
                    > paid attention to the road when driving, we wouldn't need that one),
                    > financial predictions, photo copiers (at least some). Just a few
                    > examples.
                    >
                    >
                    >
                    >
                    >
                    >
                    >
                    > --- In artificialintelligencegroup@yahoogroups.com, "Sharmontime
                    > Wonder" <sharmontime@y...> wrote:
                    > >
                    > > None has commented abt my views. Let alone sayin good, not even one
                    > > said wrong.
                    > > If i am incorrect pls correct me.
                    > > I'll be obliged.
                    > > Thanking You.
                    > > Sharmontime.W.
                    > >
                    > >
                    > >
                    > >
                    > > --- In artificialintelligencegroup@yahoogroups.com, Sharmontime
                    > > Wonder <sharmontime@y...> wrote:
                    > > >
                    > > > the neural networks are to be trained from a, b,c....(for
                    > > whatever the net is to be used for.)
                    > > > then, it is given some situations to act wherein it is suppose
                    > > to display its intelligence.
                    > > > this should be like a child is born, and you teach him the
                    > > basics, and he starts to think. whenever he is wrong, he's either
                    > > corrected by elders or learns that its fire, and i should not touch
                    > > it.
                    > > > this is what i want to simulate.
                    > > >
                    > > >
                    > > > the first sentence i mentioned :....(for whatever the net is to
                    > > be used for.), i meant that presently, i believe neural nets are
                    > > trained for some specific purposes.
                    > > > this is what i said i believe.
                    > > >
                    > > > i have not read much. i have always been fascinated by the
                    > terms
                    > > AI, NN and also robotics.
                    > > >
                    > > > i have been thinking various ways of simulating a brain. this
                    > > simulation of the brain and the nervous system is what i understand
                    > > to be the neural networks.
                    > > >
                    > > > training the net is nothing but imbibing (artificial)
                    > > intelligence in it.
                    > > > and embedding this system in electronic circuits and machines
                    > > and etc., is robotics.
                    > > >
                    > > > people interested in further discussions are welcome.
                    > > > please feel free to correct me. i am very much open to all.
                    > > > all sorts of criticisms are welcome.
                    > > >
                    > > > thanks
                    > > >
                    > > > regards
                    > > > - sharmontime
                    > > >
                    > > >
                    > > > Sathish kumar <psathishdl@y...> wrote:
                    > > > How neural network can achive artificial intelligence?
                    > > > jest by connecting neuron do you think you can achive
                    > > intelligence.
                    > > >
                    > > > do neuron processing the information ?
                    > > >
                    > > >
                    > > > Send instant messages to your online friends
                    > > http://in.messenger.yahoo.com
                    > > >
                    > > > [Non-text portions of this message have been removed]
                    > > >
                    > > >
                    > > >
                    > > >
                    > > >
                    > > > SPONSORED LINKS
                    > > > Computer science college Computer science distance
                    > > learning Computer science Artificial intelligence software
                    > > Artificial intelligence in business Artificial intelligence and
                    > > expert system
                    > > >
                    > > > ---------------------------------
                    > > > YAHOO! GROUPS LINKS
                    > > >
                    > > >
                    > > > Visit your group "artificialintelligencegroup" on the web.
                    > > >
                    > > > To unsubscribe from this group, send an email to:
                    > > > artificialintelligencegroup-unsubscribe@yahoogroups.com
                    > > >
                    > > > Your use of Yahoo! Groups is subject to the Yahoo! Terms of
                    > > Service.
                    > > >
                    > > >
                    > > > ---------------------------------
                    > > >
                    > > >
                    > > >
                    > > >
                    > > >
                    > > >
                    > > >
                    > > > ---------------------------------
                    > > > Yahoo! for Good - Make a difference this year.
                    > > >
                    > > > [Non-text portions of this message have been removed]
                    > > >
                    > >
                    >
                    >
                    >
                    >
                    >
                    > ------------------------------
                    > YAHOO! GROUPS LINKS
                    >
                    >
                    > - Visit your group "artificialintelligencegroup<http://groups.yahoo.com/group/artificialintelligencegroup>"
                    > on the web.
                    >
                    > - To unsubscribe from this group, send an email to:
                    > artificialintelligencegroup-unsubscribe@yahoogroups.com<artificialintelligencegroup-unsubscribe@yahoogroups.com?subject=Unsubscribe>
                    >
                    > - Your use of Yahoo! Groups is subject to the Yahoo! Terms of
                    > Service <http://docs.yahoo.com/info/terms/>.
                    >
                    >
                    > ------------------------------
                    >



                    --
                    Don't giveup


                    [Non-text portions of this message have been removed]
                  • predictorx
                    What about defining the fuzzy sets Mountain , Tropical and Desert , based on temperature? Any temperature could be converted to a set of 3 set
                    Message 9 of 15 , Jan 21, 2006
                    • 0 Attachment
                      What about defining the fuzzy sets "Mountain", "Tropical" and
                      "Desert", based on temperature? Any temperature could be converted to
                      a set of 3 set memberships, which can be used as weights in a weighted
                      average. See these diagrams for examples of fuzzy temperature sets:

                      http://www.ainewsletter.com/newsletters/2004_02_fuzzy2.gif
                      http://www.css.sfu.ca/update/vol8/fuzzy-diagram.gif
                      http://www.dmitry-kazakov.de/ada/ling_1.jpg


                      -Will Dwinnell
                      http://will.dwinnell.com




                      Muhammad Shoaib sehgal <thesehgal@y...> wrote:
                      > I am looking for the solution of a fundamental data
                      > fusion problem ....
                      >
                      > I have three sensors A,B,C and each sensor measures
                      > temperature in degree Centigrade.
                      >
                      > Sensor A performs well in mountanious region, B
                      > performs better than A and C in Tropical areas while C
                      > has a better performance in case of deserts.
                      >
                      > Now, I want to fuse the information with wague idea
                      > about the land type ... I am measuring the
                      > temperature using all three sensors, say for
                      > instance, show foloowing temperature:
                      >
                      > A: 20 C
                      > B: 22.5 C
                      > C: 21.9 C
                      >
                      > Now I want to fuse the information which brings simple
                      > linear equation
                      >
                      > Temperature = w1.A + w2.B + w3.C
                      >
                      > How do I adjust the weights w1,w2,w3?
                    • Bret Smith
                      You might find this interesting to read: Los Altos Hills, CA—Neural Systems Corporation (NSC) has developed trainable digital logic that can perform neural
                      Message 10 of 15 , Feb 13, 2006
                      • 0 Attachment
                        You might find this interesting to read:

                        Los Altos Hills, CA�Neural Systems Corporation (NSC) has developed
                        trainable digital logic that can perform neural network-like functions
                        using circuitry that is typically several orders of magnitude less
                        complex than that required for equivalent conventional neural networks.
                        NSC also has developed error correction algorithms for digital
                        communication channels and hard disk drives that can achieve bit rates
                        (number of bits per second) at the Shannon channel capacity and at signal
                        to noise ratios typical of cable channels and hard disk drive read
                        channels. It is known that bit rates cannot exceed the Shannon channel
                        capacity.

                        Neural networks are devices or computer algorithms that can be trained
                        using known examples to recognize and identify objects or conditions.
                        Conventional networks are composed of units that mimic biological neurons
                        (hence the name neural network). Many units are combined and trained to
                        recognize objects or changing conditions. Their uses have ranged from the
                        recognition of printed letters to the reconfiguration of aircraft
                        autopilots.

                        However, because of their complexity, they have not found widespread use
                        in the application of artificial intelligence (AI) to consumer
                        electronics. They are complex because for each recognition a computer
                        must perform many additions and multiplications to implement each unit in
                        the network

                        In contrast, trainable digital logic does not use arithmetic to make the
                        same decisions. The logic can be constructed from conventional
                        programmable logic devices such as field programmable gate arrays. In
                        addition to the great reduction in the circuit complexity, an enormous
                        increase in speed is possible. Thus, the construction of a general
                        purpose single chip trainable recognition device capable of making one
                        million decisions a second is possible. It is predicted that these
                        devices will eventually allow the use of advanced AI in consumer
                        products.

                        Methods for attaining bit rates in low digital-to-noise ratios digital
                        communication channels that equal the Shannon channel capacity (the
                        highest possible rates) have been known for approximately 13 years. These
                        methods are used, for example, in NASA�s deep space networks. However,
                        these methods do not achieve Shannon channel capacity at the higher
                        signal-to-noise ratios that are typical of earthbound channels such as
                        DSL and other broadband links.

                        Under a grant from the Advanced Technology Program of the National
                        Institute of Standards and Technology, NSC has demonstrated methods that
                        do achieve the capacity for both digital communications channels and hard
                        disk drives at signal-to-noise ratios that are usually found in these
                        devices. It is expected that communication channel rates and disk drive
                        densities can be increased by up to 25 percent using the new NSC methods.

                        NSC is a small privately held corporation headquartered in Los Altos
                        Hills, CA.

                        Contact at: http://www.neuralsyscorp.com .


                        ----- Original Message -----
                        From: "Dave Faulkner"
                        To: artificialintelligencegroup@yahoogroups.com
                        Subject: Re: [Artificial Intelligence Group] How neural network can
                        achive artificial intelligence?
                        Date: Tue, 3 Jan 2006 14:35:50 -0500


                        Here it is, in a nutshell, as I understand it --

                        Neural networks , like other mathematical constructions such as
                        fuzzy logic networks, radial basis functions, Fourier series, etc.
                        belong to a class of functions that are Universal Approximation
                        Functions.

                        UFAs are functions like NN that can mimic -- to some degree of
                        approximation -- many other well behaved functions, and some
                        not so well behaved functions.

                        So, an NN can approximate a polynomial, or a sinusoid, or a linear
                        switching function if you tweak the parameters properly. The hard
                        part is knowing what these tweaking parameter values are, because
                        they dictate whether the NN looks like a polynomial or a circle
                        or whatever.

                        And if you don't know the function, that's ok too because you can
                        make the NN resemble any set of XY graphable data -- to some
                        degree of resemblance -- simply be having the actual data. The
                        Data can be measured or derived or whatever; the NN can be
                        made to "fit" the data.

                        Key in understanding how this can possibly be "intelligence" is
                        realizing
                        that the NN can also mimic very complicated logic.

                        For example, Do I need to go grocery shopping?

                        I need to go grocery shopping if --

                        - there's no food in the house.
                        - I can pay for it
                        - the car works
                        - etc....

                        So you can set up the NN to evaluate the situation and give you
                        and "intelligent" response --

                        NN (Grocery shop) == (No food) AND (Money) AND (Working Car)

                        Now imagine 100,000 of such NNs with input to each other and the
                        ability to change each others logic parameters and you'll begin to
                        imagine a small working brain (like an insects). Most usable and
                        trainable
                        NNs are small enough to tackle just one problem, so they resemble
                        large equations. But the functions that they can mimic can be
                        large and complex, and so appear "intelligent" in the very
                        specialized context that they were created in.







                        ----- Original Message -----
                        From: "Pesky Bee"
                        To:
                        Sent: Tuesday, January 03, 2006 6:28 AM
                        Subject: Re: [Artificial Intelligence Group] How neural network can
                        achive artificial intelligence?


                        > Connectionists think that intelligence can be achieved by
                        > neural networks not because what happens with an individual
                        > neuron, but because what emerges from the operation of
                        > a large number of them. Even if a the information processing
                        > of a single neuron hardly can be considered "intelligent",
                        > the same might not be true about a group of neurons. Of
                        > course, artificial intelligence using artificial neural
                        > networks is still a hypothesis, although results so far
                        > have been promising.
                        >
                        > *PB*
                        >
                        >
                        >
                        > ----- Original Message ----- From: "Sathish kumar"
                        > To:
                        > Sent: Monday, January 02, 2006 2:07 AM
                        > Subject: [Artificial Intelligence Group] How neural network can
                        > achive artificial intelligence?
                        >
                        >
                        >> How neural network can achive artificial intelligence?
                        >> jest by connecting neuron do you think you can achive
                        intelligence.
                        >>
                        >> do neuron processing the information ?
                        >>
                        >>
                        >> Send instant messages to your online friends
                        http://in.messenger.yahoo.com
                        >>
                        >> [Non-text portions of this message have been removed]
                        >>
                        >>
                        >>
                        >>
                        >>
                        >>
                        >> Yahoo! Groups Links
                        >>
                        >>
                        >>
                        >>
                        >>
                        >>
                        >
                        >
                        >
                        >
                        >
                        >
                        > Yahoo! Groups Links
                        >
                        >
                        >
                        >
                        >



                        Yahoo! Groups Links




                        --
                        ___________________________________________________
                        Play 100s of games for FREE! http://games.mail.com/



                        [Non-text portions of this message have been removed]
                      Your message has been successfully submitted and would be delivered to recipients shortly.