[mlpack-git] [mlpack] Create a function to visualize the features learned by sparse autoencoder (#465)

[stereomatchingkiss]

>How long does it take to train the sparse autoencoder on the data from display_network.m, and how big is that data

If using golden model, I think we can skip the training process but feed the data MaximalInputs directly.

    arma::mat parameters(5, 5);
    parameters<<.....//insert data same as the one feed to matlab
    arma::mat goldenData(2,2);
    goldenData<<.....//insert the data same as the output of matlab

    arma::mat output;
    MaximalInputs(parameters, output);
    //compare the output and goldenData

>Can you elaborate?
Sure, the check I want to perform is 

    auto const squareRows = static_cast<arma::uword>(std::sqrt(paramTemp.n_rows));
    if(squareRows * squareRows != paramTemp.n_cols)
    {
        throw std::range_error("The visible size of the parameters must be integer greater than 0");
    }



---
Reply to this email directly or view it on GitHub:
https://github.com/mlpack/mlpack/pull/465#issuecomment-152037601
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.cc.gatech.edu/pipermail/mlpack-git/attachments/20151028/e230c155/attachment.html>