Hello everyone,

I am Shubham Shukla a junior student at Indian Institute of Information
Technology and Management, Gwalior. I work on Deep Learning, Computer
Vision and Web Development.
I am going to be working on Inside the Black Box project where we try to
provide control and information to the user during training, especially
through wrappers.

The major goals of my project would be:

1) Premature Stopping:
It is possible at the moment to stop/pause/resume a Machine. We will be
implementing special methods to all iterative algorithms that are
meaningful in that context to us.
This can provide more information/control about how a model is doing during
training.

2) Implementation of progress bar all over shogun:
The progress bar was implemented in 2017 by my mentor and this year I am
going to implement it in iterative algorithms. This will be helpful for our
first objective because we will
be diving into the exact places in the code base that would be affected by
our first objective.

3) Refactoring and cleanup:
A lot of code in Shogun is getting old. We are going to rewrite code in a
better and more efficient way, like rewriting unit tests and examples. This
will also help us to explore our domain for above two goals.

4) Refining the TensorBoard Integration:
We are going to do some more work on ShogunBoard particularly on feature
Visualizations of different data types (images, audio, etc). This
willinvolve tweaking the ParameterObserver Framework.

Apart from this I will also be working on using the new factory API in meta
examples.
This is going to be a very interesting summer for me and I am really happy
to be here. The work I would be doing is definitely impactful and in fact
this is what attracted me to shogun in the first place.
I will be posting weekly updates at my blog https://shubham808.github.io/

Thank you

Reply via email to