[mlpack] GSOC-2020 | Project Proposal | Application Based Machine Learning Course using MLPack

[Shivam Behl]

Hello Mentors,

I am Shivam Behl, 3rd year undergraduate at Thapar University, India.
I want to propose an "application-based Machine Learning course for MLPack"
for GSOC 2020. I have seen a previous GSOC Project which dealt with making
machine learning course for Humanity students under the organization -
RedHen Labs. Please review this short proposal and tell if it is GSOC level.

I have spent last week learning and understanding MLPack library. During
this period I realized that while learning MLPack, one is quite on your its
for a lot of time. Documentation is good, but many parts are not clearly
explained and one has to refer the source code to understand a lot of
stuff. This is is not a big deal for a seasoned programmer fluent in C++,
but can be a cause of trouble for new entrants.

I believe having a machine learning course based on MLPack can help boost
the user base of the library and help find applications in new niches.

I propose to make the course around these points -
1. Basic Machine Learning Implementations using C++ and Python Wrapper.
2. Practical applications using Mini Projects.
3. Understanding CodeBase

Work will be in three phases, as follows -
-- Phase 1 --
The intuition of Machine Learning Algorithm.
Application using ML Pack.
This pattern will be followed for all the Algorithms implemented in MLPack.

-- Phase 2 --
Hands-on Tutorials on Implementing MLPack on free to use Datasets.
Interesting visualizations and comparisons with other libraries.

-- Phase 3 --
Silent C++ features used in MLPack codebase.
Basics of how the code is implemented.
Appendix of Armadillo, Numpy Stack, etc.

The Course will involve slides, pdf for detail text and code snippets.

Presently, I am thinking of the following hands-on Tutorials to be a part
of the course -
--- Stock Market analysis and building algo-trading model using basic ML
Algorithms.
--- Style transfer using ANN
will add more soon.

Please let me know if this project is suitable for GSOC and what changes I
should make to the overall plan.

Regards,
Shivam Behl
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://knife.lugatgt.org/pipermail/mlpack/attachments/20200319/be23ed97/attachment-0001.htm>