[mlpack] GSoC - MatLab / Python Bindings

[Russell Caughey]

Hi,

I am very interested in working with you this summer in GSoC. I would like
to work to create bindings for mlpack in either MatLab or Python. My
interest in machine learning runs deep and I will bring great dedication
and enthusiasm to this project.

(basic information)

Russell Caughey

russellcaughey at gmail.com

I received a Bachelor’s of Science in Psychology with a minor in Computer
Management Information Systems from Liberty University in 2012. I attended
classes remotely while traveling and living abroad in countries across
Europe, Asia and South America. I believe it is partly due to my
unconventional educational experience that I am drawn to the open-source
community and its potential to solve problems in education and development
across the globe.

I am currently enrolled in prerequisite courses (e.g. Calculus, Discrete
Mathematics, Linear Algebra, and Differential Equations), in anticipation
of applying for a Master’s in Computer Science program this Fall.

My interest in mlpack is to be able to learn about machine learning to
implement in educational software. I want on creating software that can
both teach and learn  and adapt to it’s users (learning styles,
disabilities, etc).

(coding/machine learning experience)

I am primarily proficient both Java/Android programming. I have worked with
MathLab, R, and other statistical software packages such as SPSS and STATA.
My experience with C++ goes back to my pre-high school years when I taught
myself the language for fun, but I will be taking taking a C/C++ course
this summer to relearn anything I’ve forgotten. I do not know about SFINAE,
CRTP or delayed evaluation but I look forward to diving into these concepts!

    I have been formally programming in my coursework for 3 years, but my
interest stems from over 10 years of computer coding for pleasure. I am
very interested in researching the intricacies of parallel programming and
machine learning. I will also be taking a data algorithms course this
summer and look forward to how it will relate to mlpack!

(competence verification "challenge questions")


In a paragraph, describe some elements of good software design.

In software design, I believe it is a good idea to organise modules that
are highly related together in a hierarchical fashion based on the
generality or specialization of the module. This is mainly for the reason
that this is the way humans think. If we want to write code that is
intuitive and accessible to other software developers, we need to write in
a way that makes sense and is human readable (/*which also requires good
commenting!*/).

Suppose that I have a program, but I don't know if it has any bugs in it.
 How would you write another program which could detect whether or not my
program had any bugs in it?

I would begin by writing a program that first detects the methods available
in the program being analyzed, and then access and run each of these
methods. I would have this meta-program keep detailed logs of every in an
out of the methods being accessed.

Describe the most interesting application of machine learning you can think
of, and then describe how you might implement it.

One of the most interesting applications of machine learning, to me, is the
use of software that is learning it’s user. If the software was designed to
teach a student Calculus, it could measure mean levels in all of the
foundational skills needed (e.g. to take a derivative of y=x^3 / 14y). If
this was crossed analyzed with how the student learned best (visually,
auditory, working problem sets) with a large enough data-set, the software
could choose from a database of options in which to direct the student to
the next learning objective with the goal of achieving the highest maximum
efficiency in learning.

 Thanks for your consideration and I look forward to hearing back from you!

Best,
Russell
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.cc.gatech.edu/pipermail/mlpack/attachments/20130428/1d00036e/attachment-0002.html>