[mlpack-git] [mlpack] [Proposal]Develop a scalable Finetune class to fine tune the paramters of deep network (#458)

[stereomatchingkiss]

I do some change on the api of FineTuneGradient, this api may provide better optimization codes

    class SoftmaxFineTune
    {
     public:
      template<typename T>
      static void LastGradient(arma::mat const &input,
                               arma::mat const &weights,
                               T const &model,
                               arma::mat &gradient)
        {
            gradient = (weights.t() * model.Probabilities()) /
                    static_cast<double>(input.n_cols);
            gradient = gradient % (input % (1 - input));
        }

        static void Gradient(arma::mat const &input,
                             arma::mat const &weight,
                             arma::mat const &deriv,
                             arma::mat &output)
        {
            output = (weight.t() * deriv) % (input % (1 - input));
        }
    };

---
Reply to this email directly or view it on GitHub:
https://github.com/mlpack/mlpack/issues/458#issuecomment-146073692
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.cc.gatech.edu/pipermail/mlpack-git/attachments/20151006/53f81788/attachment.html>