Improvisation and Implementation of ANN Modules

In this project, my goal is to update the existing ANN facilities within mlpack and the deliverable is an improved ANN module. This includes adding many different functionalities, including layers like Pixel Shuffle layer and Spatial Dropout layer among many others. I have also proposed to add extended support for reduction facilities to be integrated into many different loss functions to bring to the table a degree of flexibility offered in other major machine learning libraries. I will also be building on the work done in a GSOC project last year, the mlpack-Tensorflow translator and mainly I will be updating it with all the latest changes in the source code to support new functionalities which have become a part of mlpack since last year. And finally, I have also proposed to begin work on improving the documentation of all layers, by adding inline examples within the header descriptions to improve usability for a wider range of end-users.