Genetic Algorithms and Evolutionary Computation
Adam Marczyk
http://www.talkorigins.org/faqs/genalg/genalg.html
This is a final draft, so no public links have been made just yet. Anyone
have any comments or criticisms?
--
"We have loved the stars too fondly | a.a. #2001
to be fearful of the night." | http://www.ebonmusings.org
--Tombstone epitaph of | e-mail: ebonmuse!hotmail.com
two amateur astronomers, | ICQ: 8777843
quoted in Carl Sagan's _Cosmos_ | PGP Key ID: 0x5C66F737
----------------------------------------------------------------------
Ken Shaw
Adam Marczyk wrote:
> At long last:
>
> http://www.talkorigins.org/faqs/genalg/genalg.html
>
> This is a final draft, so no public links have been made just yet. Anyone
> have any comments or criticisms?
>
It seems fundamentally correct on the details of what a GA is and how
they are used to solve problems. The use of GA to bolster the ToE is
something I'm not too sure about.
Ken
xtmprs...@erols.com
If evolution can't possibly work, then genetic algorithms
won't work either.
Harlequin
news:6Xiic.15447$_o3.4...@bgtnsc05-news.ops.worldnet.att.net:
Well we need to keep things in context. GAs are certainly
not evidence for evolution (Douglas Theobald's "29+ Evidences for
Evolution" is the approprate for that), nor is it evidence
that the proposed mechanisms of evolution actually did historically
did happen (just because it could happen does not mean that it
actually does). But it does refute the "common sense" notion/creationist
argument that the random varation combined with selection cannot produce
novel "designs." That what Darwin proposed can be put to
work on real world problems and is not a mere "higgledly
piggledy" hypothesis is important.
Besides, the whole subject of genetic algorithms simply damning to
the "no new information" claims of antievolutionist in general
and well as the ID movement (especially Dembski) in particular.
This actually brings sometime up. Adam in the article mentions
as one of the "limitations" of GAs is that they "cheat" by
solving a problem that the programers did not intend to but
strictly speaking "solved" what they were asked to. (Sort
of the letter and the spirit of the law.) He gives the example
of an oscillating signal generator being the target and instead
a radio was evolved that picked-up oscillating signals.
This is actually a great example of how evolution in nature
actually does work: take advantage of what is in the
enviroment. But is does show another disadvantage to selecting
GA's only in the computer. This experiment actually had
each mutant circuit made in the real world. If the oscillating
signal had been computed rather than measured this solution
would not have evolved. Thus GAs can have an heuristic value
in explaining the ToE as well.
--
Anti-spam: replace "usenet" with "harlequin2"
Tired of creationist political victories undermining science education?
Sign up for low-volume moderated email updates on creationist
activity. A list for every U.S. state and two Canadian provinces:
http://www.aibs.org/mailing-lists/the_aibs-ncse_evolution_list_server.html
Join the National Center for Science Education. See
http://www.ncseweb.org/membership.asp to join online (or offline).
Adam Marczyk
I hope I haven't conveyed the impression that the success of genetic
algorithms on engineering and design problems proves that evolution
was the historical process by which living things diversified on this
planet. That would obviously be fallacious - we need to look at the
evidence from nature to determine if biological evolution has actually
happened, and I think Talk.Origins does a very good job of surveying
that evidence in different articles. The point of mine was a bit
different, as I hoped to show from the conclusion:
"As incredible as it may seem, evolution works."
The point of my article was to demonstrate that evolutionary processes
actually can produce complex structures and new information, in
contradiction to creationist claims. I didn't intend to show that
evolution did happen in nature, but I did intend to support the claim
that it *could* have.
xtmprs...@erols.com
Another example: If that radio transmission had stopped that
species of circuit would have become extinct.
Don1
> Ken Shaw <non...@your.biz> wrote in message news:<6Xiic.15447$_o3.4...@bgtnsc05-news.ops.worldnet.att.net>...
> > Adam Marczyk wrote:
> >
> > > At long last:
> > >
> > > http://www.talkorigins.org/faqs/genalg/genalg.html
> > >
> > > This is a final draft, so no public links have been made just yet. Anyone
> > > have any comments or criticisms?
> > >
> >
> > It seems fundamentally correct on the details of what a GA is and how
> > they are used to solve problems. The use of GA to bolster the ToE is
> > something I'm not too sure about.
>
> I hope I haven't conveyed the impression that the success of genetic
> algorithms on engineering and design problems proves that evolution
> was the historical process by which living things diversified on this
> planet. That would obviously be fallacious - we need to look at the
> evidence from nature to determine if biological evolution has actually
> happened, and I think Talk.Origins does a very good job of surveying
> that evidence in different articles. The point of mine was a bit
> different, as I hoped to show from the conclusion:
>
> "As incredible as it may seem, evolution works."
>
> The point of my article was to demonstrate that evolutionary processes
> actually can produce complex structures and new information, in
> contradiction to creationist claims. I didn't intend to show that
> evolution did happen in nature, but I did intend to support the claim
> that it *could* have.
It is great that you put this article up. I have been anticipating it.
However, when I go to the page, I get the following:
The XML page cannot be displayed
Cannot view XML input using XSL style sheet. Please correct the
error and then click the Refresh button, or try again later.
--------------------------------------------------------------------------------
End tag 'ul' does not match the start tag 'li'. Error processing
resource 'http://www.talkorigins.org/faqs/genalg/genalg.html'.
Line 170,
Position 3
</ul>
--^
Are you in the middle of editing something?
Regards,
Don
Adam Marczyk
news:56db0569.04042...@posting.google.com...
I haven't modified it since I put it up. The page loads fine for me - is
anyone else having this problem?
Jon Fleming
It loads fine in Opera 7.23 or Mozilla 1.6, but in IE 6.0.2800.1106 I
get the same error he reported.
<http://validator.w3.org/check?uri=http%3A%2F%2Fwww.talkorigins.org%2Ffaqs%2Fgenalg%2Fgenalg.html>
lists 415 errors in that document.
--
Replace nospam with group to email
Jon Fleming
<jo...@fleming-nospam.com> wrote:
><http://validator.w3.org/check?uri=http%3A%2F%2Fwww.talkorigins.org%2Ffaqs%2Fgenalg%2Fgenalg.html>
>lists 415 errors in that document.
A lot of them are failure to close ... I don't think there are any
</li> elements in there. A lot of them are improper nesting like
<h2>...<p></p></h2>.
Harlequin
news:3bml80puqgl3ouqh4...@4ax.com:
I just did a few minutes light work with HTML-Kit. The page
now validates according to the W3C validation Service and
HTML Tidy. Adam, in case of need, there is a back up
on the server. There are still a variety of minor HTML chores
to do but I will let Adam do any other changes that he needs
to do first.
(HTML-Kit is just great. Just hit F9 and make one click
and all those failures to properly close the tags just
went away.)
In any event, any problem that any browser has due because
of invalid coding should be fixed.
Don1
Thanks I can see it now.
I have some questions, though, about the html/xsl/xml.
It is true that different browsers do render incorrect html differently.
More often than not, IE requires html tags to be closed, while somehow
NetScape figures it out. I remember a case of a FORM tag not being closed.
NetScape still was able to put in a button at the end of the form and the
form action still worked. In IE the form was not usable.
But I received an error message about XSL. The application of the xml to
the xsl occurs server-side and not client-side, correct? So exactly what
happened here? Was there some intermediate request/response between the
browser and server? There has to be an explanation of how different browser
rendering of html can produce an xsl parsing error message...
Harlequin
news:56db0569.04042...@posting.google.com:
> Harlequin <use...@cox.net> wrote in message
> news:<Xns94D5C0EC43335u...@68.12.19.6>...
>> Jon Fleming <jo...@fleming-nospam.com> wrote in
>> news:3bml80puqgl3ouqh4...@4ax.com:
>>
>> > On Sat, 24 Apr 2004 21:09:50 +0000 (UTC), Jon Fleming
>> > <jo...@fleming-nospam.com> wrote:
>> >
>> >><http://validator.w3.org/check?uri=http%3A%2F%2Fwww.talkorigins.org%
>> >>2Ff aqs%2Fgenalg%2Fgenalg.html> lists 415 errors in that document.
Not true. Badly constructed HTML is rampant through out the web.
Though bad HTML risks random bad interpretation by browsers, most
of the time they figure out the intent correctly. The very fact
that you can read most documents is proof of this.
> while somehow NetScape figures it out. I remember a case of a FORM tag
> not being closed. NetScape still was able to put in a button at the
> end of the form and the form action still worked. In IE the form was
> not usable.
And sometimes IE figures it out and Netscape does not.
> But I received an error message about XSL.
Your browser must have a serious bug since XSL was never in
any way, shape, or form used by Adam's document. There simply
was not declaration that an XSL stylesheet existed and nor
were any XSL tags used.
> The application of the xml
> to the xsl occurs server-side and not client-side, correct?
It can be either way. But up-to-date browsers can use
an XSL "style sheet" to convert an XML document that uses
it to a displayable page. But one can use scripts server
side to do it as well.
> So exactly
> what happened here? Was there some intermediate request/response
> between the browser and server? There has to be an explanation of how
> different browser rendering of html can produce an xsl parsing error
> message...
If your browser really thought it was seeing XSL then it
would have required the pages to be valid since XML is requires
documents to be well formed. I wonder if you browser failed to
read the !DOCTYPE declaration of transitional XHTML which
is an XML version of HTML in that it is supposed to be
well formed. Admittedly any document with such a
declaration is supposed to close all tags which Adam's document
did not, but usually browsers
will still try and interpret it the best they can.
Don1
It looks like my wording was syntactically ambiguous. Let me explain
again
what I meant.
"It is true that different browsers do render incorrect html
differently. " -- What I meant was exactly what you said. When the
html
is incorrect, there is often not a standard specification for how
to
treat it. Sometimes it is rendered properly and sometimes it is
not
within any browser.
"More often than not, IE requires html tags to be closed, while
somehow NetScape figures it out." -- When I wrote "more often than
not"
I had meant that IE is generally stricter than Netscape when it
comes
to tags being closed. In other words, you read it like this:
"(More often than not, IE requires html tags to be closed), while
somehow NetScape figures it out."
I meant it like this:
"More often than not, (IE requires html tags to be closed, while
somehow NetScape figures it out.)"
Here I used ()'s as a shortcut to order of precedence in X-bar
theory.
If you still do not see the two different interpretations in the
text,
then I will post the sentence diagrams of each.
It has been my experience developing web pages that IE is in fact
stricter
than Netscape. As you stated, incorrect html can result in a poorly
rendered
web page, regardless of browser, too. In this specific case, Opera
7.23 and Mozilla 1.6 were less strict. Also IE 5.50.4522.1800 and IE
6.0.2800.1106
were both more strict.
>
> > But I received an error message about XSL.
>
> Your browser must have a serious bug since XSL was never in
> any way, shape, or form used by Adam's document. There simply
> was not declaration that an XSL stylesheet existed and nor
> were any XSL tags used.
>
I was using IE 5.50.4522.1800. Jon Fleming reported the same exact
error
message in IE 6.0.2800.1106.
I don't think that serious bug is the only possible conclusion, though
I
would wager it to be the most likely.
Another possibility would be typo in the header info (such as doctype
etc)
in the file. There could be a case-sensitive or punctuation issue,
which is
only noticed in IE because it looks for the exact standard spec. I
remember,
for example seeing a difference in the way web servers handle case and
punctuation differences in the content-type message.
> > The application of the xml
> > to the xsl occurs server-side and not client-side, correct?
>
> It can be either way. But up-to-date browsers can use
> an XSL "style sheet" to convert an XML document that uses
> it to a displayable page. But one can use scripts server
> side to do it as well.
>
> > So exactly
> > what happened here? Was there some intermediate request/response
> > between the browser and server? There has to be an explanation of how
> > different browser rendering of html can produce an xsl parsing error
> > message...
>
> If your browser really thought it was seeing XSL then it
> would have required the pages to be valid since XML is requires
> documents to be well formed. I wonder if you browser failed to
> read the !DOCTYPE declaration of transitional XHTML which
> is an XML version of HTML in that it is supposed to be
> well formed. Admittedly any document with such a
> declaration is supposed to close all tags which Adam's document
> did not, but usually browsers
> will still try and interpret it the best they can.
>
I think you are correct here. The open tags somehow impacted the
interpretation of the doctype declaration and surrounding tags.
However, I would not exclude the possibility of some mistake in/around
the
!DOCTYPE declaration. My suggestion would be to look up the standard
spec
and doublecheck the file.
Adam Marczyk
Since the page now validates as XHTML 1.0, and since there seem to be no
serious critiques of the content, I'm going to go ahead and put up public
links later tonight. Does anyone else have anything important to point out
that I've overlooked or gotten wrong?
steve_h
Don1 wrote:
>
> Thanks I can see it now.
>
> I have some questions, though, about the html/xsl/xml.
>
> It is true that different browsers do render incorrect html differently.
> More often than not, IE requires html tags to be closed, while somehow
> NetScape figures it out. I remember a case of a FORM tag not being closed.
> NetScape still was able to put in a button at the end of the form and the
> form action still worked. In IE the form was not usable.
>
> But I received an error message about XSL. The application of the xml to
> the xsl occurs server-side and not client-side, correct? So exactly what
> happened here? Was there some intermediate request/response between the
> browser and server? There has to be an explanation of how different browser
> rendering of html can produce an xsl parsing error message...
The client side stuff is in XHTML not just plain HTML (see the
DOCTYPE declaration at the top). XHTML should be "well formed" like
XML - For instance if you look at the source the line breaks
<br> (which do not need to be matched by </br> in HTML, appear
as <br />. The browsers that were _not_ complaining had
it wrong in this case.
Steve
Harlequin
[snip]
I have had not had the experience of IE not rendering invalid
(X)HTML for being invalid. And it is not for the lack of
viewing them in IE. What is really strange is that I am
using IE 6.0.2800.1106 (though actually it says "1106CO").
I also count about 15 "update versions" listed in "About".
> I don't think that serious bug is the only possible conclusion, though
> I would wager it to be the most likely.
>
> Another possibility would be typo in the header info (such as doctype
> etc) in the file. There could be a case-sensitive or punctuation issue,
> which is only noticed in IE because it looks for the exact standard spec.
> I remember, for example seeing a difference in the way web servers handle
> case and punctuation differences in the content-type message.
[snip]
>> > So exactly
>> > what happened here? Was there some intermediate request/response
>> > between the browser and server? There has to be an explanation of
>> > how different browser rendering of html can produce an xsl parsing
>> > error message...
>>
>> If your browser really thought it was seeing XSL then it
>> would have required the pages to be valid since XML is requires
>> documents to be well formed. I wonder if you browser failed to
>> read the !DOCTYPE declaration of transitional XHTML which
>> is an XML version of HTML in that it is supposed to be
>> well formed. Admittedly any document with such a
>> declaration is supposed to close all tags which Adam's document
>> did not, but usually browsers
>> will still try and interpret it the best they can.
>
> I think you are correct here. The open tags somehow impacted the
> interpretation of the doctype declaration and surrounding tags.
>
> However, I would not exclude the possibility of some mistake in/around
> the !DOCTYPE declaration. My suggestion would be to look up the standard
> spec and doublecheck the file.
The !DOCTYPE declaration was okay (or rather would have
been okay if the coding had complied with the declaration).
I have seen it validate MANY times from the W3C validator and other
validators. (Adam was using an old version of the Archive
template probably because he had been working on the document
for quite some time.) I have also verified the !DOCTYPE
against what is given by the W3C.
In any event, you have shown another reason to be strict
about validation.
I knew there was a reason why I send
anything I work on to the W3C Validation Service.
(Indeed I programmed a keyboard shortcut for it rather
to eliminate the need to go through HTML-Kit's menu
each time.)
Harlequin
news:c6ip8a$4b5$1...@atlas.ip-plus.net:
I absolutely agree that XHTML should be well-formed
(and indeed MUST be well-formed to be valid).
Indeed, I always send XHTML code I do to the W3C validation
Service. But it still find it a bit strange that a browser
would not try to render the document anyways. After all
most composers produce code that is in some way faulty often
to the degree that one has to wonder if the people behind
it even know HTML. Lets face it, 90% of the sites on the web do
not comply with any standards whatsoever. While it might
be a purist thing to say that documents should comply with
the standards with no mistakes, the reality is that is unlikely
in the real world and not trying rendering them might put the
browser at a competive disadvantage. Maybe a good idea whould
be to do what IE does for Javascript errors: put icon on
bottom of the window to indicate error is present in the page.
What is particularly strange to me is that a browser
created by Microsoft would object to invalid code
since Microsoft certainly is not exactly known for
complying with the W3C standards.
Ken Shaw
I'm still puzzled by this whole problem. I viewed this document before
the folks who reported the error. I used IE 6.0.2800.1106 and it worked
fine. IME IE rarely chokes on unclosed tags of the sort found in the
original version of this document. The only condition I know a about
that will really give IE fits is unclosed TD and TR tags inside a deeply
nested set of tables.
Ken
Thomas H. Faller
Nah... It would have shifted up a couple of Hz and become a photocell...
;-)
Tom Faller
Kari Tikkanen
: At long last:
: http://www.talkorigins.org/faqs/genalg/genalg.html
: This is a final draft, so no public links have been made just yet. Anyone
: have any comments or criticisms?
There is something missing. Of course no list can be perfect but now also
Finnish elevator/lift company Kone seems to be using genetic
algorithms for optimized routing of elevator cars etc..:
http://www.kone.com/en/main/0,,content=33602&topelem_id=32818&navielem_id=32836,00.html
"The KONE Alta™ elevator design is developed for super-tall office
buildings up to 500 meters (1,600 ft) in height. These elevators can
transport up to 70 passengers and reach speeds up to 17 meters per second
(3,400 ft/min). This is possible thanks to powerful double-rotor versions
of the KONE EcoDisc® hoisting machines.
KONE Alta™ control systems employ genetic algorithms for optimized
routing of elevator cars using fuzzy logic and artificial intelligence
for specific control tasks. Special KONE 'SilentCar' technology applied
on both single and high capacity 'double-deck' cars ensures a smooth and
quiet ride."
Also
http://www.cs.hut.fi/~psu/VK94/node41.html
Also
In Finnish Magazine "Helsingin Sanomat" 22 April 1995 page D2
GA simulation results of Kone elevators was introduced in science and
environment pages (just computer simulations of GA in 1995):
"In experiments (done in University of Vaasa) showed that GA-control has
shortened waiting delay even 12% compared with older waiting delay times.
Evolution was allowed going on quite a long time, an hour. In elevators
GA should make decisions less than in half seconds.
Even small increment of efficiency is remarkable advantage for funding
company Kone in elevator markets. Thousands impatient button pressers
are waiting at the moment in hundreds of skycrapers" (translation to
English by me)
->Finnish elevation factory/company (KONE) is going to apply
genetic algoriths and evolution principles to elevator
controlling computer programs
->simulations, made in University of Vaasa, have shown that
waiting delay shortened 12 % in some cases compared with
traditional programs.
->shows that principles of evolution work - even in computer
-http://www.student.oulu.fi/~ktikkane/eEVOTOD.html (point A.)
Another point. From my memory. One engineer here has claimed that GA
does not produce really new but uses existing parts. Well, like giving
impression that there are not mutations..
Regards,
Mr. Kari A. Tikkanen, Finland.
BTW. Which is more evil from creationists (inside scycrapers) point of
view: 1) evolution algorithms inside or 2) fundies outside ?
--
http://www.student.oulu.fi/~ktikkane
eNo
> At long last:
>
> http://www.talkorigins.org/faqs/genalg/genalg.html
>
> This is a final draft, so no public links have been made just yet. Anyone
> have any comments or criticisms?
Adam, I won't comment on the technical details of your article, as I'm
no expert in this field. But as a hardware/firmware/software engineer, I
found your article extremely informative and thought provoking. Here are
some ramblings that crossed my mind, in no particular order, and with no
necessary validity implied (more along the "what if?" kind):
1) Could Natural GA (NGA), as observed in nature, be in itself an
"intelligent" process? Just thinking outloud, one could say that:
a) GA shows how NGA could give the appearance (echoes of "Blind
Watchmaker") of intelligence where a mere random process, governed by
fitness parameters is at work, or
b) GA/NGA shows a level of intelligence unlike that described by ID
proponents, i.e., deprived of a final goal, and yet can arrive at fairly
complex and innovative solutions as a result of adaptive iterations.
IOW, is it possible that natural selection *itself* is a pre-designed
algorithm? Or conversely, could GA (or NGA, by extension) succeed in the
absence of prescribed conditions (i.e., the laws of nature/physics,
natural selection, the fitness criteria), which in the case of
artificial GA, were defined by intelligent agents?
Whatever we say about Natural Selection/NGA and artificial GAs, most of
us would agree they produce some very clever solutions that may appear
strange to us until we dig deeper. Now we can ponder ad infinitum
whether that "cleverness" is real or a figment of our imagination.
--
øĪš°`°šĪø,,,,øĪš°`°šĪø,,,,øĪš°`°šĪø,,,,øĪš°`°šĪøĪš°`°šĪø,,,,øĪš
eNo
"If you can't go fast, go long."
øĪš°`°šĪø,,,,øĪš°`°šĪø,,,,øĪš°`°šĪø,,,,øĪš°`°šĪøĪš°`°šĪø,,,,øĪš
Adam Marczyk
news:8hAjc.8$ei...@dfw-service2.ext.ray.com...
> Adam Marczyk wrote:
>
>> At long last:
>>
>> http://www.talkorigins.org/faqs/genalg/genalg.html
>>
>> This is a final draft, so no public links have been made just yet.
>> Anyone have any comments or criticisms?
>
> Adam, I won't comment on the technical details of your article, as I'm
> no expert in this field. But as a hardware/firmware/software engineer, I
> found your article extremely informative and thought provoking. Here are
> some ramblings that crossed my mind, in no particular order, and with no
> necessary validity implied (more along the "what if?" kind):
>
> 1) Could Natural GA (NGA), as observed in nature, be in itself an
> "intelligent" process? Just thinking outloud, one could say that:
> a) GA shows how NGA could give the appearance (echoes of "Blind
> Watchmaker") of intelligence where a mere random process, governed by
> fitness parameters is at work, or
> b) GA/NGA shows a level of intelligence unlike that described by ID
> proponents, i.e., deprived of a final goal, and yet can arrive at fairly
> complex and innovative solutions as a result of adaptive iterations.
> IOW, is it possible that natural selection *itself* is a pre-designed
> algorithm?
That's actually the argument Dembski makes in No Free Lunch, although he
doesn't seem to recognize it. Really it's just the argument from
cosmological fine-tuning stated in different terms.
> Or conversely, could GA (or NGA, by extension) succeed in the
> absence of prescribed conditions (i.e., the laws of nature/physics,
> natural selection, the fitness criteria), which in the case of
> artificial GA, were defined by intelligent agents?
I'm not sure what you're saying here. Are you asking if evolution could
occur if there were no laws of physics?
> Whatever we say about Natural Selection/NGA and artificial GAs, most of
> us would agree they produce some very clever solutions that may appear
> strange to us until we dig deeper. Now we can ponder ad infinitum
> whether that "cleverness" is real or a figment of our imagination.
--
Adam Marczyk
news:c6mja9$mqh$1...@ousrvr3.oulu.fi...
> Adam Marczyk <ebon...@deletethis.hotmail.com> wrote:
>> At long last:
>
>> http://www.talkorigins.org/faqs/genalg/genalg.html
>
>> This is a final draft, so no public links have been made just yet.
>> Anyone have any comments or criticisms?
>
> There is something missing. Of course no list can be perfect but now also
> Finnish elevator/lift company Kone seems to be using genetic
> algorithms for optimized routing of elevator cars etc..:
>
>
http://www.kone.com/en/main/0,,content=33602&topelem_id=32818&navielem_id=32836,00.html
> "The KONE Alta™ elevator design is developed for super-tall office
> buildings up to 500 meters (1,600 ft) in height. These elevators can
> transport up to 70 passengers and reach speeds up to 17 meters per
> second (3,400 ft/min). This is possible thanks to powerful double-rotor
> versions of the KONE EcoDisc® hoisting machines.
>
> KONE Alta™ control systems employ genetic algorithms for optimized
> routing of elevator cars using fuzzy logic and artificial intelligence
> for specific control tasks. Special KONE 'SilentCar' technology applied
> on both single and high capacity 'double-deck' cars ensures a smooth and
> quiet ride."
>
> Also
> http://www.cs.hut.fi/~psu/VK94/node41.html
>
> Also
> In Finnish Magazine "Helsingin Sanomat" 22 April 1995 page D2
I'd be happy to add this to my article if you tell me the name of the
author so I can cite it fully. What does Helsingin Sanomat mean in English?
> GA simulation results of Kone elevators was introduced in science and
> environment pages (just computer simulations of GA in 1995):
>
> "In experiments (done in University of Vaasa) showed that GA-control has
> shortened waiting delay even 12% compared with older waiting delay
> times. Evolution was allowed going on quite a long time, an hour. In
> elevators GA should make decisions less than in half seconds.
You said this was done in 1995. Have there been any updates since? Does
there actually exist a GA that can come up with these more efficient
solutions quickly enough to compete with traditional algorithms?
eNo
Thanks for asking. To be honest, I'm not sure what I'm asking either. ;)
Perhaps something along the lines of: Would we have NGA if the laws of
physics and other physical parameters weren't as they are, and what does
this indicate regarding the underlying "intelligence" of genetic
algorithms in general?
Kari Tikkanen
: >
: > Also
: > In Finnish Magazine "Helsingin Sanomat" 22 April 1995 page D2
: I'd be happy to add this to my article if you tell me the name of the
: author so I can cite it fully.
Author is science writer Timo Paukku. He often writes articles in
science+environment pages in Helsingin Sanomat.
Here I give full referee:
"Timo Paukku: Evoluutio ohjaa hissinkin taipaletta, Helsingin Sanomat,
tiede.ymparisto, 22 Apr 1995, D2"
In that article Timo Paukku interviewed researcher Juha Herajarvi (Juha
Herajärvi in scandic characters) from university of Vaasa.
(Headline "Evoluutio ohjaa hissinkin taipaletta" would be
in english :"Evolution controls also movement/travel of elevator")
> What does Helsingin Sanomat mean in English?
Literally ? 'Helsinki's messages'. But I don't recommend using that..
Perhaps I used wrong term 'Magazine'. It is quite high quality daily
newspaper like "New York Times". This "Helsingin Sanomat" is Finland's
greatest newspaper, name refers to capital city Helsinki.
http://www.helsinginsanomat.fi/english/
: > GA simulation results of Kone elevators was introduced in science and
: > environment pages (just computer simulations of GA in 1995):
: >
: > "In experiments (done in University of Vaasa) showed that GA-control has
: > shortened waiting delay even 12% compared with older waiting delay
: > times. Evolution was allowed going on quite a long time, an hour. In
: > elevators GA should make decisions less than in half seconds.
: You said this was done in 1995. Have there been any updates since? Does
: there actually exist a GA that can come up with these more efficient
: solutions quickly enough to compete with traditional algorithms?
In fact I don't know updates , at least I've not seen any more news about
Kone's GA in Helsingin Sanomat or any science popular magazines here. It
could be business secret, too(?).
As far as I know: In those 1995 experiments their used simulations of GA
and simulations of conventional algorithms. All that elevator systems
stuff with virtual customs etc was simulated and compared inside computer.
And betterment with GA was 12% shorter times.
So newest news about KONE using GA I got googling this week like here:
http://www.hightechfinland.com/2002/newmaterials-processes/sivu.php?id=kone&listby=
"The Traffic Master 9900 GA control system ensures minimum passenger
waiting and flight times. The system incorporates a genetic algorithm
with multi-target-optimising capability, employing artificial
intelligence and fuzzy logic to carry out support tasks. "
They tell that Kone Alta systems (which includes GA) for very tall have
been tested in Tytyri 333 meter high limestone mine in Finland. And it
seems now be in real use:
"The wold's first KONE Alta™ elevators were installed in
Aurora Place, Sydney in 2000"
http://www.hightechfinland.com/2003/newmaterialsprocess/transport/kone.html
I think some world's largest skycrapers in Asia are going to have Kone
Alta(tm) systems. They have preliminary contract. So they gonna have GA's
too.
From Helsinki University of Technology pages I found this:
http://www.cs.hut.fi/~psu/VK94/node41.html
---------------------------------------------------------------------
"GATE - Industrial Applications of Genetic Algorithms
...
...
-------------------------------------------------------
The system developed for Kone turned out to be a success.
-------------------------------------------------------
After some technical operating system problems, the distributed GA-based
system developed for Neste engineering is now in their production
use. The most valuable outcome for the company was a distributed
parameter estimation system that is faster and more reliable than
their former system.
The system developed for Kone turned out to be a success solving the
elevator control problem better than the current methods and lead to a
larger project that aims at developing a new GA-based elevator group
controller. "
-----------------------------------------------------------------------
There I found link to scientific articles:
-----------------------------------------------------------------------
[1]
Jarmo T. Alander. An indexed bibliography of genetic algorithms: Years
1957-1993. Art of CAD Ltd., Vaasa (Finland), 1994. (over 3000 GA
references).
[2]
Jarmo T. Alander, editor. Proceedings of the Second Finnish Workshop on
Genetic Algorithms and their Applications, Vaasa (Finland), 16.-18. March
1994. University of Vaasa, Department of Information Technology and
Industrial Management.
[3]
Jarmo T. Alander, Antti Autere, Janne Mäntykoski, and Kari I. Keskinen.
Distributed genetic algorithm for fitting of model parameters
of chemical reaction kinetics. In Alander[2], pages 115-126.
[4]
Jarmo T. Alander, Tapio Tyni, and Jari Ylinen. Optimizing elevator group
control parameters using distributed genetic algorithm. In Alander[2],
pages 105-113.
[5]
Jarmo T. Alander and Jari Ylinen. Application of genetic algorithms to
the determination of protein 3D structure from NMR spectroscopy data. In
The First Electronic Computational Chemistry Conference, page Paper 49,
Internet / www, 1994. (represented at the internet hypermedia conference,
now in review for publication in a CD-ROM).
[6]
Jarmo T. Alander, Jari Ylinen, and Tapio Tyni. Optimizing elevator group
control parameters using distributed genetic algorithms. In
D. W. Pearson, N. C. Steele, and R. F. Albrecht, editors, Artificial
Neural nets and Genetic Algorithms, Proceedings of International
Conference (ICANNGA95), pages 400-403, Al`es (France), 18.-21. April
1995. Springer-Verlag, Wien
-------------------------------------------------------------
At least [4] and [6] are related to GA's in elevators.
Another link gives:
"Jarmo Alander .. Acting professor, production automation
University of Vaasa, Department of Information Technology and
Industrial Management"
Shortly:
GA's appeared to be more efficient than conventional algorithms in
elevator simulations in 90's, and
now in 21th century they are in succesful commercial use by Kone
Corporation.
--
Kari Tikkanen ! . . -#- ! b ! begin
FIN-90550 OULU ! ! ! I = / f(x)dx ! s:=s+Eq(i);
FINLAND ! . . Vega ! a ! end
http://www.student.oulu.fi/~ktikkane
Stanley Friesen
>Literally ? 'Helsinki's messages'. But I don't recommend using that..
>Perhaps I used wrong term 'Magazine'.
Perhaps "Helsinki News"? Or "Helsinki Reports"?
The peace of God be with you.
Stanley Friesen
Walter Bushell
"Adam Marczyk" <ebon...@deletethis.hotmail.com> wrote:
> http://www.talkorigins.org/faqs/genalg/genalg.html
wow!