Re: Working on nnet package

[Julien Bect]

Le 29/12/2016 à 11:37, Francesco Faccio a écrit :

I started adding a few tests in dividerand.m and I realized that it didn't work properly (probably Matlab has changed a lot of features in the last few years). 

Although Matlab documentation is not very exhaustive, I tried to make the code more compatible, I added some tests and improved the documentation. 

You can find the commit here:
https://bitbucket.org/Francesco_Faccio/octave-nnet/commits/a064c3b83c994d07aaecc20131765c35e89549f0

Am I working in the right way? If so, I'm going to do the same for the other functions in the package.

Hi Francesco,

I am not familiar with the nnet package, but I can give you a few general pieces of advice...

1) Do not top-post on this mailing list---i.e., write your answer *below* the previous message, not *above*.

2) Do not expect a lot of answers about this old package.  Development by the original author, Michael Schmid, stopped in 2010 [1, 2] and it became unmaintained, which indicate a lack of interest in the OF community.  Someone [3] wrote a message about nnet and neural networks in general in 2015 but nobody answered...

3) Before starting to work on bringing nnet back to life, have you considered other potentially useful ways to spend your time and contribute to OF?  There are many well-developed open source packages for neural networks, and it would be interesting to have bindings for them in Octave, in the form of OF packages.

4) For instance, Søren Hauberg wrote a Octave bindings for FANN [4], which were later modified by Hans Fugal [5].  I don't know how mature this package is.  Some adaptations might be necessary to bring the package up to date with FANN 2.1 and Octave 4.2.0.  Such a package would be an interesting addition to OF.

5) Other popular NN packages worth considering include Torch [6], TensorFlow [7] (Google's NN library), NetLab [8] (a little old perhaps)...

I am not saying that you should give up your project of bringing nnet back to life, if this is what you want to do.  Just that there are other options worth considering...

@++
Julien


[1] https://sourceforge.net/p/octave/nnet/commit_browser
[2] http://plexso.com/octave/index.html
[3] http://octave.1599824.n4.nabble.com/Neural-Network-Package-needed-td4668734.html
[4] http://leenissen.dk/fann/wp/
[5] https://github.com/fugalh/octave-fann
[6] http://torch.ch/
[7] https://www.tensorflow.org/
[8] http://www.aston.ac.uk/eas/research/groups/ncrg/resources/netlab/news/