[mlpack] GSoC'21 Work Report

[Roshan Swain]

Hi,
I created a work summary for GSoC'21. There were a lot of blockers like
long training time, jupyter kernel crashes(always behaves strangely), and
other personal challenges, but still somehow made it till now. I have
compiled my final works below:

1. Implementation of Linear Regression for California Housing:
 This includes CPP and Python Notebooks with cool visualizations and
evaluation metrics.

2. Implementation of Generative Adversarial Networks for MNIST:
This covers usage of GAN class with documentation and comments on Generator
and Discriminator Architecture. This is a major headache coz it is taking a
lot of time to train and thus very time consuming to optimize.

3. Implementation of GAN for Anime Faces Generation:
The idea was to showcase usage of GAN for creation of detailed anime
faces(RGB) but considering the training time, we altered it a bit. We are
gonna translate Pytorch model weights and convert them into mlpack weights.
This is in progress and hopefully should be completed by this week.

Link to GSoC work report:
https://github.com/swaingotnochill/gsoc-work-report
[ I will be making updates to the report too accordingly].

On a final note, I am very grateful to the mlpack community, especially
Ryan and Marcus for helping me get started with mlpack initially and my
mentors Marcus and Kartik, they always helped me whenever I was in a
pinch(I loved the zoom meetings :D). Also, thanks to David for helping out
a lot in this time period. Throughout this period, I enjoyed contributing
to mlpack and will continue to do so.
Thank You.

Regards,
Roshan Swain
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://knife.lugatgt.org/pipermail/mlpack/attachments/20210820/b1d0bf4a/attachment.htm>