Dear Roland,

In my opinion the directions in Issam's deep learning proposal are a
bit better suited for scikit-learn.  Our estimators are supposed to be
black-box and as general as possible with sensible defaults.  I don't
know to what extent recurrent nets can be implemented in such a way.

Could you discuss a bit how recurrent networks with long-short term
memory would fit in scikit-learn and in what ways they would be
useful?

As for the Kohonen SOM, for some reason academics in our country
*love* them, but they had enough time to prove themselves useful and
apparently have not.
There was a failed attempt to merge this a while back:
https://github.com/scikit-learn/scikit-learn/pull/39

In my opinion autoencoders are exciting as a scikit-learn
contribution. A good design should be figured out to allow sharing of
low-level code with the MLP.

Yours,
Vlad

On Sat, Apr 20, 2013 at 12:43 AM, Roland Szabo <[email protected]> wrote:
> Hi!
>
> I'm a 2nd year student at Babes-Bolyai university in Cluj. I am interested
> in contributing to scikit-learn and in participating in GSOC.
>
> I am doing my bachelor's thesis about neural networks and I would to
> implement some of the most commonly used ones in scikit-learn.
>
> I know there are two pull requests about multi-layer perceptrons (one from
> Lars and one from Hannes) and one for Restricted Boltzmann Machines. I read
> that Andreas would like to merge the RBM as soon as possible, but the MLP
> pull request still has plenty of work left to do.
>
> Besides this I would like to implement autoencoders, Kohonen Self-Organizing
> Maps[1] and Long Short Term Memory[2]
>
> I have already contributed a three small fixes to scikit-learn (and if you
> have any other issues with which I can help I would be happy to do so).
>
> What do you think?
>
> --
> Roland
> http://rolisz.ro/
>
> [1]
> http://www.eicstes.org/EICSTES_PDF/PAPERS/The%20Self-Organizing%20Map%20(Kohonen).pdf
> [2] http://www.cs.umd.edu/~dmonner/papers/nn2012.pdf
>
>
> ------------------------------------------------------------------------------
> Precog is a next-generation analytics platform capable of advanced
> analytics on semi-structured data. The platform includes APIs for building
> apps and a phenomenal toolset for data science. Developers can use
> our toolset for easy data analysis & visualization. Get a free account!
> http://www2.precog.com/precogplatform/slashdotnewsletter
> _______________________________________________
> Scikit-learn-general mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
>

------------------------------------------------------------------------------
Try New Relic Now & We'll Send You this Cool Shirt
New Relic is the only SaaS-based application performance monitoring service 
that delivers powerful full stack analytics. Optimize and monitor your
browser, app, & servers with just a few lines of code. Try New Relic
and get this awesome Nerd Life shirt! http://p.sf.net/sfu/newrelic_d2d_apr
_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general

Reply via email to