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

Re: How neural network can achive artificial intelligence?

Expand Messages
  • 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 1 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 2 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 3 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 4 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.