Le 21 mars 2012 04:55, David Warde-Farley <[email protected]> a écrit :
> On 2012-03-20, at 9:16 PM, Rami Al-Rfou' wrote:
>
>> Hi All,
>>
>> I think Torch7 and Theano are fast and powerful libraries that would be nice 
>> to take advantage of them.
>
> They're also rather heavy dependencies.
>
> In this case, since I think scikit-learn would be pretty happy with a 
> relatively basic implementation, there may be some C code in Torch7 that is 
> well-isolated enough to bundle with scikit-learn, but then again, there may 
> not be.

Indeed. I think torch7 benefits a lot from using Lua as a scripting
language but that is definitely not a runtime dependency nor a
cultural dependency we would want to add to a project such as
scikit-learn.

As for theano, there is already the pylearn2 project that addresses this scope:

  http://deeplearning.net/software/pylearn2/

If we are to add implementation for some neural nets to the project I
would rather have it implemented in pure cython without any further
dependencies and providing less flexibility on the structure of the
networks and the list of hyperparameters that any of those other
project would.

If your goal is to do research on deep learning, you should definitely
invest some time in theano + pylearn2 or torch7. If you just want to
have a "standard" feed forward MLP implementation to compare it
against other scikit-learn models such as RBF kernel SVM or random
forests, then a simple cython module would be enough.

-- 
Olivier
http://twitter.com/ogrisel - http://github.com/ogrisel

------------------------------------------------------------------------------
This SF email is sponsosred by:
Try Windows Azure free for 90 days Click Here 
http://p.sf.net/sfu/sfd2d-msazure
_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general

Reply via email to