Beefy Boxes and Bandwidth Generously Provided by pair Networks Bob
Think about Loose Coupling
 
PerlMonks  

Re^3: AI::NNEasy to setup fast a Neural Network using just Perl and XS.

by g0n (Priest)
on Jan 16, 2005 at 14:32 UTC ( #422619=note: print w/ replies, xml ) Need Help??


in reply to Re^2: AI::NNEasy to setup fast a Neural Network using just Perl and XS.
in thread AI::NNEasy to setup fast a Neural Network using just Perl and XS.

OK, I've uploaded AI-NNFlex-0.11 to CPAN, with support for datasets a little bit like the Xerion approach. It makes the UI quite a bit friendlier - if you look at ./examples/xor_with_datasets.pl you'll see what I mean. PNG support has been put into ::draw and the lesioning method has been implemented. You can now damage the network probabilistically on a network-wide, layer or node basis.

I've cleaned up some of the nastiest perldoc sections aswell.

Quick question - I haven't done any work on the XS issue. I've never encountered XS before - does it have any prerequisites? Like a C compiler?

Charles.


Comment on Re^3: AI::NNEasy to setup fast a Neural Network using just Perl and XS.
Re^4: AI::NNEasy to setup fast a Neural Network using just Perl and XS.
by gmpassos (Priest) on Jan 16, 2005 at 15:30 UTC
    XS is like another mini language. First you need a C compiler, actually you need the same compiler used to compile your Perl. So, on Linux is easy since is always with GCC, but on Win32 you will need Microsoft Visual C++ 6+ (not free) if you got Perl from Active State, or MingW, that is GCC for Win32.

    Than you need to learn how Perl works with C and the resources for that (perlapi). So, take a look in the docs: perlapi, perlguts, perlxstut. All of that you can find better at http://www.perldoc.com.

    If you already know C, I think that the best way to you make XS functions with Perl is getting as examaple the sub{C} methods that you can find in the sources of AI::NNEasy. Because this functions have a Perl and a C version, so you can compare them and understand better what the C macros does. Also the Class::HPLOO syntax is easy like sugar, since you only need to write the function, all the rest Class::HPLOO will write for you, converting to Inline::C, than Inline will convert to XS, and from XS is with Perl. So, the code below is 100% complete, you just need to have a C compiler well installed:

    use Class::HPLOO ; class Foo { sub[C] int add( int x , int y ) { int ret = x + y ; return ret ; } }
    Now you just need to run the code above, and the rest is automatically.

    Other thing, your module is inside AI::NNEasy starting from the class AI::NNEasy::NN. So, if you move this class (and the sub classes sinde AI::NNEasy::NN::*) to your project you will have your modules with the resources that I have added to it. I let it separated to be more easy to you to get back the work that I did, so be free to use it. ;-P

    Graciliano M. P.
    "Creativity is the expression of liberty".

      Thanks for that.

      It'll take a bit of time to explore this. I'm not sure I want to use HPLoo, since I intended NNFlex to be usable as a fairly simple teaching framework. The XS & HPLoo code is quite a bit more complex than the pure perl!

        "...fairly simple teaching framework."

        Yes, is much more faster to code in pure Perl, specially if your main purpose is to make a flexible NN. But I saw that even for simple problems a NN needs speed, since if we build wrong the NN configurations for the target problem it will use a lot of CPU and time in the learning fase. Even a simple "tic tac toe" game, that I have used as a 1st real test for NNEasy, uses a lot of time to learn a big amount of inputs, so I started to add the XS support what make possible to learn fast the inputs and really play. In other words, NN always use a lot of CPU, so we can't run way from the C codes. About Class::HPLOO, for me is the fastest way to add the C support for the functions that really need that.

        Graciliano M. P.
        "Creativity is the expression of liberty".

Log In?
Username:
Password:

What's my password?
Create A New User
Node Status?
node history
Node Type: note [id://422619]
help
Chatterbox?
and the web crawler heard nothing...

How do I use this? | Other CB clients
Other Users?
Others musing on the Monastery: (5)
As of 2014-04-21 02:57 GMT
Sections?
Information?
Find Nodes?
Leftovers?
    Voting Booth?

    April first is:







    Results (489 votes), past polls