MathGroup Archive 2007

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: Neural networks with mathematica

  • To: mathgroup at smc.vnet.net
  • Subject: [mg83554] Re: Neural networks with mathematica
  • From: Jean-Marc Gulliet <jeanmarc.gulliet at gmail.com>
  • Date: Thu, 22 Nov 2007 06:37:43 -0500 (EST)
  • Organization: The Open University, Milton Keynes, UK
  • References: <fhu7mu$7cv$1@smc.vnet.net> <fi0ovm$670$1@smc.vnet.net> <fi1304$col$1@smc.vnet.net>

dinodeblasio at gmail.com wrote:
> On 21 Nov, 10:07, Jean-Marc Gulliet <jeanmarc.gull... at gmail.com>
> wrote:
>> dinodebla... at gmail.com wrote:
>>> i would like to run one ART1 network, to this purpose i have found the
>>> seguent code on some article:
>> It may be interesting to state what article it is, from whom, and where
>> it has been published (in a scholarly journal, an edited book,
>> proceedings of a conference, general audience magazine, web site, etc.)
>> so readers of this newsgroup may also read it and better understand what
>> the problem is.
>>
>>> "these are some variables"
>>> a1_ = 1
>>> b1_ = 1.5
>>> c1_ = 5
>>> d1_ = 0.9
>>> el_ = 3
>>> del1_ = 0.2
>>> del2_ = 0.1
>>> "then i would run the code written below, but i have some problem and
>>> it doesnt work, it never finish to calculate. You can find the
>>> Mathematica package below at:http://www.esnips.com/web/dinodeblasio/
>> <snip>
>>
>> This package as well as some others are available on MathSource at
>>
>> http://library.wolfram.com/infocenter/Books/3485/
>>
>> This page contains material that supplement James A. Freeman's book,
>> "Simulating Neural Networks with Mathematica", Addison-Wesley, 1994.
>>
>> So, why not using the link to the official web site for the code?
>>
>> (By the way I could not help noticing that your web page contains a link
>> to an illegal copy of a copyrighted book for another very well known
>> numerical software used mainly by engineers).
>>
>> Thus, to get meaningful answers, it would be netter to use and post
>> official sources for software and copyrighted material in addition to
>> clearly state what your problem is (the directions you followed from the
>> article or book, the expected or original author result(s), the
>> result(s) you got from your trials, a comparison or analysis of the
>> discrepancies, any error or warning messages, the version of Mathematica
>> you are using, etc.).
>>
>> Regards,
>> --
>> Jean-Marc
> 
> Dear Jean-Marc, thanks for your quick help.
> By the way i am not in complete agreement with you for some reason.
> 1)The text i found on internet that I called "article" is "Reti
> neurali: ipotetiche forme di vita artificiale o realistici calcolatori
> alternativi?", you can find on internet the manuscript. I did not say
> exactly what it was because is in Italian and since I guess that not
> everybody understands Italian here i did not mention it.

You have a point here. Still, it would have been interesting to 
summarize the article or a lest the main points that are relevant to 
your experiment with Mathematica.

> 2)I found the package from internet before I found the book"
> Simulating neural network with mathematica" that i have now.

The point I raised (I admit I should have phrase it more explicitly) is 
depending of the source where you got the package, you cannot be sure 
that the code had not been modified in any way (except if the site can 
be trusted and some up to date documentation is provided). That is why 
it is usually better to use the original source.

> 3) I guess that this is a forum, free, for free development of the
> culture. and I think that if i gave you a link to the code and there
> was also one pdf. of one book this is not a big deal. You have never
> downloaded one mp3 from internet for free?

My comment was not intended to blame you but to emphasize (and I might I 
failed to do so) that although this newsgroup is public, it still have a 
code of conduct (it is moderated) and follows a proper Net Etiquette 
(which, among many other things, objects promoting or advertising 
illegal material).

> 4)Since i work in university my software and books are all genuine of
> course, and the law in this country states that if you have a book you
> can make a copy for yourself.(because for example you can write on it,
> the same is with pdf files)

Fair enough.

> 5)Since you found a copy (not illegal, i just made an error to place
> it in the public folder, now is not public anymore) of a book on an
> other software; you could not help me? what is that, I should use only
> mathematica and not an other software?

Not at all. My point was not to blame or reprimand you. If you read the 
last paragraph of my previous reply, you will see that I am asking for 
more information about your problem.

> 6)if you are not able or you cant answer to questions for any reason,
> why you argue on what i should or not should do, and just dont write
> anything?This would avoid me to waste time.

I was willing to help, but to do so, I inquired more information about 
the issue(s) you faced. (Again read the last paragraph.) Yet you have 
not provided more clues about your problem. (In my case, if the issues 
are related to Mathematica, I can certainly help. OTOH, if the issues 
are about neural network and the like, I will not be of great help for 
it is neither my domain of expertise nor a field of personal interest.) 
The fact that I have been the only person to reply to your message (at 
least at the date of this message) should be a good hint that something 
was incorrect or missing.

> Regards.
> 

By the way, I have reply (in another post) to your original post about 
some possible syntax errors and I reported what I got by running the 
code you posted on my system. Hope this will help.

Regards,
-- 
Jean-Marc



  • Prev by Date: Re: memory release problem in mathematica6.0
  • Next by Date: Export figures with tooltips
  • Previous by thread: Re: Neural networks with mathematica
  • Next by thread: Re: Neural networks with mathematica