Incorrect, you look at it from the point at which
you are at now. What if you watched you would find
that the child builds randomly based on rules made by
that child. As things begin to work more one begins to
build a cache of processes that work as one of the many
chess models I've built does. <br><br>NO I don't have
any to give away I wish I did.<br><br>This building
of technique is simply a matter of fuzzy guessing.
"It has a 90% chance of
raining."<br><br>Spider_plant9

Ahh.. Pathoneurology nice field. Now, as for how
the brain function it is a rather simplistic computer
system based on threshold logic.<br><br>A neuron fires
down its axon which is recieved by the dendrite of the
neighboring neuron. If the neighbor recieves more pulses from
other neurons it may reach its threshold voltage
(roughly -500mlv) where the neuron that has reached
threshold there by fires down its axon and continues the
logic.<br><br>Looking at how the harder ware of the body works it is
much like many microprocessors working together to
manipulate the IO.<br><br>Now, in torets (I'm guessing
haven't read this far yet) the threshold could be set too
low or the amount of axon signal being recieved to
too great. The first would be simple to change by
changing the balance of Na, K, and Ca ions in the nervous
system. The other problem would involve rewiring. Any
volunteers to rewire a network of pentiums at the transistor
level.<br><br>Spider_plant9

hi!<br> I'm an eleventh standard student and I'm
preparing for a national level interview. I want to know as
much about artificial intelligence as possible in as
brief a duration as I have (20 days). Can anybody tell
me where I can find concise information about
AI.<br><br>Thanks in Advance<br>Parijat

Hi, Parijat!<br>Welcome to the World of AI!<br>If
you have some time to look through books or even to
read them in 20 days, I recommend ***Elain Rich's
"Artificial Intelligence"*** (though it is from the 80-s and
does not cover neural networks, I think). I just love
it. Another one that has been much spoken about it
Nilssen's new introductory book, though i haven't read it
yet.<br>As for online stuff, try *John McCarthy's Home
Page*:<br><a href=http://www-formal.stanford.edu/jmc/frames.html
target=new>http://www-formal.stanford.edu/jmc/frames.html</a><br>Joe McCarthy is
one of the pioneers of AI and has a
very lucid style.<br>Another pioneer's site: (Marvin
Minsky)<br><a href=http://www.ai.mit.edu/people/minsky/minsky.html
target=new>http://www.ai.mit.edu/people/minsky/minsky.html</a><br>And also look
at:<br><a href=http://ai.iit.nrc.ca/ai_point.html
target=new>http://ai.iit.nrc.ca/ai_point.html</a> - A good collection of
resources and
information.<br>And, if there is more time left, look at the Yahoo! AI
Category:<br><a
href=http://dir.yahoo.com/Science/Computer_Science/Artificial_Intelligence/
target=new>http://dir.yahoo.com/Science/Computer_Science/Artificial_Intelligence\
/</a><br><br>Hope it helps,<br>Felix.

FREE STUFF<br><br>> The Robot Club of Traverse
City is for girls and boys, women and
men,<br>anyone<br>> with an interest in robots. Come to the meeting
to meet other people<br>> interested in robots
and see what they are doing. Help others with
their<br>> robots. Get help with your robot.<br><br>The
Robot Club of Traverse City meets the second Tuesday of
each month.<br><br>> The next meeting of the Robot
Club of Traverse City will be 9/12/00 at
6<br>pm<br>> at the Manufacturing Technology Academy 715 E.
Front St. Traverse City,<br>MI<br>>
49686<br>><br>> MTA is in NMC's Maritime Academy Building at 715
E. Front Street (just<br>east<br>> of the Holiday
Inn) . MTA is in the southeast corner of the
building.<br>You<br>> should park in the lot directly south of the
building (opposite the bay)<br>and<br>> enter through
the glass door on the right (east) as you face the
building.<br>> MTA is the first set of door on your left.
There's a sign.<br>><br>> Program item 1: MTA will
demonstrate their student's award winning robot<br>>
simulation software.<br>><br>> Program item 2: You are
welcome to show what you have been working
on.<br>><br>> Program item 3: Parts swap and free robot stuff
give away.<br>><br>><br>> See you
there.<br>><br>> Paul<br>><br>> Paul F. Grayson - Chief
Engineer<br>>AMERICAN INDUSTRIAL MAGIC [inventors for hire ]<br>>
1892 Pinewood Ave.<br>> Traverse City, MI
49684-9022<br>> (231) 946-0187, FAX (231) 946-1122<br>>
pgrayson@...<br>>

Even if the intelligence making program prints
out all the moves, we still won't know what what
intelligence is, or even how it is produced. We have all of
Beethoven's notes, but we still cannot explain his genius, or
exactly what he did. It will be the same with programs
created with autonomous agents. We can read away all we
want, but we'll only be guessing as to what is taking
place.

Your Comparison of Beethoven's notes is not the
same thing. Beethoven's notes are the reults, not the
process. What I am talking about recording is every move
made from the inside.: a large project, to be
sure.<br><br>It wouldn't, in itself, tell us where the
intelligence is, but it would give us a roadmap, which is more
than we have right now.

Actually, no, the composition is the process, the
result is listening to the music. It is the relationship
of the notes to each other that creates music, and
we have to hear that. It would be the same with an
intelligence program. We might guess at how the relationships
between the codes produce intelligence, but we won't be
sure. We have theories about music, but the theory does
not help us compose it. Such composition relies on
the ineffible arrangement in the composer's mind,
that even he or she does not understand.

I still don't think that you're understanding
me.<br><br>I'm talking about every computation the computer
makes, every change of memory. We don't have that with
Beethoven's notes, because we don't have what was going
through his head. What we would need to record is every
command executed, every change of memory. Not just the
responses it give to the outside world.<br><br>Ever heard
of hidden layers in networks?

I do understand you! I only wish you were right.
It is very annoying to me what I am saying, which is
we will never understand what intelligence is, even
though we will be the evolutionary conduit for its
progress. I did not come to this easily. I am a control
freak!<br><br>It does not matter if we record every command, every
change of memory. Why? Because we are dealing with
emergence, and we can analyze and interpret until we are
blue in the face, but we still will not be able to
understand how all these minute changes manage to do their
thing. The process of natural selection will always
remain a mystery.<br><br>And this just pisses the hell
out of me.

When studying Neural networks one can easily see
how they can be trained to recognise certain objects
or carry out certain behaviors.<br><br>Where does
the training come from?<br>What decides to carry out
a behavior?<br><br>It is likley that computers will
have an advantage over biological neural nets because
computers can dynamically create and allocate resources to
various networks so that each function gets exactly as
much as neccessary. While the brain has a rather
static structure. <br><br>When studying senses one
learns that a bug looks tasty to a frog's eye. (Yes
frogs have "bug Detectors" among the cells in their
eyes).<br><br>How does this perception of an object become
connected to the behavior of slurping a bug down?<br><br>=\

Good point. I guess I didn't realize that, or
didn't want to. <sigh> I agree with you about not
being able to predict, or even understand, emergent
prperties. That does suck.<br><br>Hey! Something strange
happened in this discussion. You changed my opinion.
People's mind's and opinions don't change often.

>It is likley that computers will have an
<br>>advantage over biological neural nets because
<br>>computers can dynamically create and allocate
<br>>resources to various networks so that each <br>>function
gets exactly as much as neccessary. <br>>While the
brain has a rather static structure. <br><br>Here's a
question: could limited memory actually Contribute to
intelligence?

You cheered me immensely with your reply.  We should come up with a bumper
stick, "Emergence Sucks"

I've seen in Stanford Encyclopedia of Philosophy
(<a href=http://plato.stanford.edu/entries/church-turing/
target=new>http://plato.stanford.edu/entries/church-turing/</a>) a sentence
stating that there are abstract
machines already constructed that prove the falsity of
Church-Turing thesis (or Church thesis, as it's called in
Mendelson's Math. Logic). Church-Turing thesis states that
anything that is computable (by some
machine/computer/brain) can be computed by a Turing machine (which is
equivalent to other formalizations of the concept of
algorithm, such as Markov's formalization, etc.). Now, this
was a surprise for me. I can't imagine proving a
non-formal machine to be stronger than some formal
machine.<br>Does anyone know an example of such abstract machine
that computes something a Turing machine can't?<br>Or
maybe someone has any ideas on how such a machine can
be constructed?<br>This is directly connected with
the artificial intelligence problem: if our computers
(which work using algorithms -- Turing machines) can't
do something that another physical object can, then
it means there are intelligent creatures "smarter"
than computers.<br>There is some literature mentioned
in the link above, but I don't have much access to
American literature, so if anyone has some ideas on this I
will be happy to hear them.

I don't believe the problem was with emergence
and its properties. It is with our ability to predict
them. Emergence is unfoldment, or discovery. Human
history is based, to a great deal, on discoveries or
"emergence," and we can hope (or pray) to increase our ability
to predict or discern them.

Hi Miles, I believe we are talking about two
different definitions. I quote a good definition from
"Who's Afraid of Schrodinger's Cat?" Quill, William
Morrow, New York, 1997.<br><br>"A property of an entity
or complex system is said to be emergent if it
cannot be defined or explained in terms of the
properties of its parts, or if it is not reducible to these
properties and their reltations."<br><br>P.S. I used to work
as a physical therapy aid about a million years ago,
the advances are fantastic!

I guess that I just went for a simple definition.
"Emergent" as something that emerges or is emerging. If the
property can be defined or explained by some
deep-searching soul, would that mean that the property is no
longer emergent? If that is the case, then it has
emerged and is no longer emergent. How many things (or
properties) have been unexplained until someone enlightened
us?<br> By the way, I have an interest in physical
therapy. I have a web page to propose use of certain
weight equipment to strengthen the upperbody. These
machines can be used with no lower body motion. If this
interests you, click on to
<a href=http://sites.netscape.net/milescarrm/physical_therapy_modems
target=new>http://sites.netscape.net/milescarrm/physical_therapy_modems</a>

The SRC of the human thought eh. Sounds like fun
I do that with programs that build themselves
drivers so that they can figure out new
problems.<br><br>Interesting that you wouldn't just look at the SRC and all to
do that.<br><br>AR - Thought mapping (fun)

The notes are not the process of building the
sound they are the process of performing the song. The
way the note were arranged and why they were arranged
are the process of composing.<br><br>AR - I write
software and audio-ware (aka music)

Intelligence is the abitiy to make mistakes and
adapt in order not to make them again. Sentient
intellignce is the same except that it adds interaction with
experience that the intelligence hasn't experianced
directly.<br><br>SP (Now, why is that hard)

You are talking about an NN that works on the
basis of "Zener Diode". Things happen at thresholds and
hardwire programs based on these thresholds are pretty
simple. I program in this manner when it comes to making
a new program for the hell of making one.<br><br>I
usually have no particular program type in mind but I
usually end up with something interesting. I have a
library of about 20 common program snippits. These are
called my amino acids I arrange them in what ever way I
want if something useful pops out I transcribe the
order of the snippits. (There's a story here about
evolution)<br><br>SP - Yeah I actually built a program that did this
and after a while it became an avid writer of
programs and poetry.

Depends on the speed of processor and what is meant by limited memory.  Active
(RAM), ROM, or Drive space (pad and paper vs. 30Gb harddrive.)<br><br>SP

Musical notes are individual entities that
arrange themselves according to natural selection - so we
have a kind of natural selection of aesthetic
composition. Therefore, the notes and process are the same.

No. You maintain the AI prejudices of yore, using
prejuducial cultural interpretations of this mind phenomena.
In the end you will get a machine that can solve
problems, but will it whisper sweet nothings in your ear?

"How do you make the NN operate in a
Pseudo-Parallel fashion? You can only have so many threads of
execution and even so, the processor can only execute on
instruction at a time. A brain is massively parallel and
signals are sent simultaneously across the network. How
do you do that on a computer?"<br><br>Typically, the
state of each of the parallel units is maintained
seperately, and the serial computer operates on them one at a
time. This is how virtually all abstract parallelism is
handled by conventional computers.

Dear friends of intelligence, especially the
artificial one...! I am wondering why people, in many clubs
and forums, talk about not knowing what exactly
intelligence is. 'Intelligence' is an English word with
certain known definition, easily found in all
dictionaries. So I suppose that it is something else, not
intelligence, that they talk about. What is it?<br><br>ali_mk.

how about this:<br><br>you can create layers that
are executed sequentially such that each neuron sends
its outputs to neurons in the next layer. Some
neurons from the Biological NN would obviously have to be
duplicated and the resulting NN could be far more complex
than the one being modeled.<br><br>I can reduce simple
nets to this schema but I am not sure that if it can
be proved that any massivley parallel network can be
reduced into the sequential execution of layers of
neurons.

Hello, I think you're simply right. M. Minsky
says that in "The Society Of Mind". We all know what
it's desired to construct, but no one knows how to
create it.<br>To accomplish with the second task, a
progressive refinement of a dictionary's definition of
"intelligence" is needed. Any refinement will be a major
solution for the problem of AI.

Musical notes do NOT arrange themselves- they are arranged in the mind of the
composer, not on the sheet of paper.