[MWForum]Re: Neural Nets
Bob Gorman
mwforum@lists.mathcats.com
Wed, 22 May 2002 12:55:15 -0400
--=====================_12272058==_.ALT
Content-Type: text/plain; charset="us-ascii"; format=flowed
At 11:30 AM 5/22/2002, Gary McCallister wrote:
> Is anyone on the forum familiar with neural nets?
Yes. I've coded them in Basic and used some of the commercial products but
that was a long time ago.
>Has anyone tried to program one in MW?
No, not yet, but I may be interested if I re-awaken my interest in Cancer
research.
My brother's battle with Multiple Myeloma is currently at a draw, the
cancer has neither grown nor shrunk for almost a year.
>I am starting to learn a little about neural nets and was thinking that it
>would be interesting and fun to program a single artificial neuron in MW.
This would be interesting and a true learning experience. The power of the
Sum of weighted inputs, some of which are feedback signals from an earlier
output, is a powerful & fascinating concept. I don't believe it has been
fully explored at all.
>Before jumping in with both feet I though I would see if I was
>re-inventing the wheel, or if anyone might have any suggestions for an
>approach.
A single neuron could be done in MW or MSWLogo, or any of the other flavors.
Once done, one might move to a multi-turtle logo such as StarLogo.
I would also be interested in the neural way we form rigid ideas under the
duress of trauma, and how that creates/maintains multiple personalities.
P.S.
Some of my earlier thoughts:
The ratio of noise to signal in AI is perhaps 1000:1 or worse. I have
tracked it for years and was, and still am sadly disappointed. On the
plus side the best reference is "The Handbook of Artificial Intelligence"
published as 3 volumes in 1982. It is a very good overview of the field.
Unfortunately, there has been little truly new ideas since then.
One magazine, that does have good explanatory articles is PC AI.
Unfortunately, the articles have been repetitious since there is little new
work being reported.
Neural Networks has been almost totally bought into by the economic
community. IMHO this is truly sad as these people, who firmly believe the
future is predicted by a cyclic projection of the past, are turning to NN
in what I believe is a true grasping at straws...
My focus is learning or growth systems, but this has not caught on either
in professional papers or practical applications.
"To create New Answers; you must ask New Questions."
- Bob Gorman
http://www.kncell.org
--=====================_12272058==_.ALT
Content-Type: text/html; charset="us-ascii"
<html>
<font size=3>At 11:30 AM 5/22/2002, Gary McCallister wrote:<br>
<blockquote type=cite cite> Is anyone on the
forum familiar with neural nets? </font></blockquote><br>
Yes. I've coded them in Basic and used some of the commercial products
but that was a long time ago.<br>
<br>
<blockquote type=cite cite><font size=3>Has anyone tried to program one
in MW? </font></blockquote><br>
No, not yet, but I may be interested if I re-awaken my interest in Cancer
research.<br>
My brother's battle with Multiple Myeloma is currently at a draw, the
cancer has neither grown nor shrunk for almost a year.<br>
<br>
<blockquote type=cite cite><font size=3>I am starting to learn a little
about neural nets and was thinking that it would be interesting and fun
to program a single artificial neuron in MW.
</font></blockquote><br>
This would be interesting and a true learning experience. The power of
the Sum of weighted inputs, some of which are feedback signals from an
earlier output, is a powerful & fascinating concept. I don't believe
it has been fully explored at all.<br>
<br>
<blockquote type=cite cite><font size=3>Before jumping in with both feet
I though I would see if I was re-inventing the wheel, or if anyone might
have any suggestions for an approach.</blockquote><br>
A single neuron could be done in MW or MSWLogo, or any of the other
flavors.<br>
Once done, one might move to a multi-turtle logo such as StarLogo.<br>
<br>
I would also be interested in the neural way we form rigid ideas under
the duress of trauma, and how that creates/maintains multiple
personalities. <br>
<br>
P.S.<br>
Some of my earlier thoughts:<br>
The ratio of noise to signal in AI is perhaps 1000:1 or worse. I have
tracked it for years and was, and still am sadly
disappointed. On the plus side the best reference is
"The Handbook of Artificial Intelligence" published as 3
volumes in 1982. It is a very good overview of the field. Unfortunately,
there has been little truly new ideas since then. <br>
<br>
One magazine, that does have good explanatory articles is PC AI.
Unfortunately, the articles have been repetitious since there is little
new work being reported.<br>
<br>
Neural Networks has been almost totally bought into by the economic
community. IMHO this is truly sad as these people, who firmly believe the
future is predicted by a cyclic projection of the past, are turning to NN
in what I believe is a true grasping at straws...<br>
<br>
My focus is learning or growth systems, but this has not caught on either
in professional papers or practical applications.<br>
</font><br>
<font size=4>"</font><font face="Times New Roman, Times" size=4>To
create
</font><font face="Times New Roman, Times" size=4 color="#FF0000">New</font><font face="Times New Roman, Times" size=4>
</font><font face="Times New Roman, Times" size=4 color="#008000"><b>Answers</font></b><font face="Times New Roman, Times" size=4>;
you must ask</font>
<font size=4 color="#FF0000"><b>New</font></b><font size=4>
</font><font size=4 color="#008000"><b>Questions</font></b><font size=4>."<br>
</font>- <font color="#FF0000">Bob</font> <font color="#008080">Gorman<br>
</font><font color="#800000"><a href="http://www.kncell.org/" eudora="autourl">http://www.kncell.org</a></font> </html>
--=====================_12272058==_.ALT--