[mlpack] Warm up challenges for 'Reinforcement Learning' and 'Essential deep learning modules'

Marcus Edel marcus.edel at fu-berlin.de
Fri Mar 3 09:55:50 EST 2017


Hello Satyaki,

welcome and thanks for getting in touch.

> Good to see mlpack in 2017’s organisation’s list. I am final year bachelor’s
> student, will be joining Carnegie Mellon University as a graduate student this
> fall with a specialisation in deep learning.  I had participated in GSoC 2015
> with JdeRobot.  I also have done an internship from CMU Robotics institute on
> deep learning, publication : https://www.ri.cmu.edu/pub_files/2016/11/cmu-ri-tr-
> Maturana.pdf

Looks like, using deconvolution network to detect cars in photographs is a
robust method that achieves good results, putting it on my reading list. Also
eager to hear what you did at JdeRobot if you like to elaborate.

> I was looking for some projects related to deep learning in GSoC 2017 and the
> two ideas in mlpack caught my attention.


The Reinforcement learning and Essential deep learning modules project has been
discussed at length on the mailing list before:

http://mlpack.org/pipermail/mlpack/2017-March/003095.html
http://mlpack.org/pipermail/mlpack/2017-March/003098.html
http://mlpack.org/pipermail/mlpack/2017-February/003087.html

http://mlpack.org/pipermail/mlpack/2017-March/003107.html
http://mlpack.org/pipermail/mlpack/2017-February/003092.html

Note that there are many more posts on this in the mailing list archive
to search for; those are only some places to get started.

> Could you let me know about any warm up challenges or issues that need to be
> patch in order to get started or before I start preparing my proposal.

There are some issues open in the Github issue tracker, and any contributions of
new techniques or efficiency improvements for existing implementations are
always welcome.

Besides, I think a good warmup task for the reinforcement learning project is to
implement a simple agent that is able to solve simple tasks, that way you learn
about reinforcement learning and at the same time dive into the codebase. The
same applies for the  Essential deep learning modules project idea if you have
an interesting model in mind that is manageable to implement in a short time I
think that would be a great first contribution.

I hope this is helpful, let us know if you have any more questions.

Thanks,
Marcus

> On 2 Mar 2017, at 18:47, Satyaki Chakraborty <satyaki.cs15 at gmail.com> wrote:
> 
> Hello developers,
> Hello Marcus,
> 
> Good to see mlpack in 2017’s organisation’s list. I am final year bachelor’s student, will be joining Carnegie Mellon University as a graduate student this fall with a specialisation in deep learning. 
> I had participated in GSoC 2015 with JdeRobot. 
> I also have done an internship from CMU Robotics institute on deep learning, publication : https://www.ri.cmu.edu/pub_files/2016/11/cmu-ri-tr-Maturana.pdf <https://www.ri.cmu.edu/pub_files/2016/11/cmu-ri-tr-Maturana.pdf>
> 
> Some of my relevant deep learning projects are:
> https://github.com/shady-cs15/tiny-slash <https://github.com/shady-cs15/tiny-slash> (generating guitar music with LSTM nets)
> https://github.com/shady-cs15/lrpr <https://github.com/shady-cs15/lrpr> learning deep representations for place recognition (submitted, under review)
> https://github.com/shady-cs15/rcnn <https://github.com/shady-cs15/rcnn> RCNNs for scene labelling
> 
> I was looking for some projects related to deep learning in GSoC 2017 and the two ideas in mlpack caught my attention.
> 1. Reinforcement learning
> 2. Essential deep learning modules
> 
> I had used mlpack previously. So I am somewhat familiar with the architecture. 
> Could you let me know about any warm up challenges or issues that need to be patch in order to get started or before I start preparing my proposal.
> 
> Best,
> Satyaki
> (http://shady-cs15.github.io/ <http://shady-cs15.github.io/>)
> _______________________________________________
> mlpack mailing list
> mlpack at lists.mlpack.org
> http://knife.lugatgt.org/cgi-bin/mailman/listinfo/mlpack

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://knife.lugatgt.org/pipermail/mlpack/attachments/20170303/147776e3/attachment.html>


More information about the mlpack mailing list