[mlpack] Regarding GSoC application

Marcus Edel marcus.edel at fu-berlin.de
Sat Mar 3 07:34:05 EST 2018


Hello Akash,

> I was occupied with the examination and assignment load so I was inactive the
> previous weeks.  I have read 'Playing Atari with Deep Reinforcement Learning'. I
> would like to share that my previous semester's project was on demonstrating how
> Deep Reinforcement Learning could be applied on Self driving cars. I had read
> this paper for the background study and it helped a lot in creating a software
> model for the project. This project involved a hardware implementation that was
> possible using ROS. Now we are preparing a paper to publish regarding the same.

That sounds great.

> I have been active in the part of reading the mailing list. I like the idea of
> creating tutorials on OpenAI gym. I would love to help in making it possible. We
> could even extend this idea not only the tutorials but to blogs that not only
> have tutorials on using OpenAI gym but also on machine learning techniques and
> algorithms, that would have beginner to advanced level tutorials on using mlpack
> for machine learning, just like scikit learn and TensorFlow. This would help
> newbies in ML who already have knowledge of C++ easily use mlpack.

Thanks, I think a good start would be to do something similar to this
https://github.com/mlpack/models/pull/5, let me know what you think.

> As suggested I would move forward with policy gradients. I know I am late, and I
> hope to catch up as fast as possible.

No worries, there is plenty of time left, the application period ends on March 27.

Thanks,
Marcus

> On 2. Mar 2018, at 16:45, Akash Shivram <akashshivram9 at gmail.com> wrote:
> 
> Sure.
>  I thought my mails were public, but as you've pointed out, they arent. I notice that I have replied to your email id instead of the mailing list.
>  I would keep in mind that my mails go public. 
> 
> Thank you 
> 
> On 02-Mar-2018 6:36 PM, "Marcus Edel" <marcus.edel at fu-berlin.de <mailto:marcus.edel at fu-berlin.de>> wrote:
> Hello Akash,
> 
> do you mind if I responde to this on the public mailing list? That way more
> people can jump in and provide input.
> 
> Best,
> Marcus
> 
>> On 1. Mar 2018, at 20:11, Akash Shivram <akashshivram9 at gmail.com <mailto:akashshivram9 at gmail.com>> wrote:
>> 
>> Hey Marcus, 
>> 
>> Thanks a lot for replying.
>> I was occupied with the examination and assignment load so I was inactive the previous weeks.
>>  I have read 'Playing Atari with Deep Reinforcement Learning'. I would like to share that my previous semester's project was on demonstrating how Deep Reinforcement Learning could be applied on Self driving cars. I had read this paper for the background study and it helped a lot in creating a software model for the project. This project involved a hardware implementation that was possible using ROS.
>> Now we are preparing a paper to publish regarding the same. 
>> 
>> I have been active in the part of reading the mailing list. I like the idea of creating tutorials on OpenAI gym. I would love to help in making it possible. We could even extend this idea not only the tutorials but to blogs that not only have tutorials on using OpenAI gym but also on machine learning techniques and algorithms, that would have beginner to advanced level tutorials on using mlpack for machine learning, just like scikit learn and TensorFlow. This would help newbies in ML who already have knowledge of C++ easily use mlpack. 
>> 
>> As suggested I would move forward with policy gradients. I know I am late, and I hope to catch up as fast as possible. 
>> Thank you
>> 
>> 
>> On 14-Feb-2018 3:52 PM, "Marcus Edel" <marcus.edel at fu-berlin.de <mailto:marcus.edel at fu-berlin.de>> wrote:
>> Hello Akash,
>> 
>> thanks for getting in touch, glad you like the project idea.
>> 
>> Getting familiar with the codebase especially
>> src/mlpack/methods/reinforcement_learning/ should be the first step, as you
>> already pointed out. Running the tests: (rl_components_test.cpp)
>> 'bin/mlpack_test -t RLComponentsTest' and (q_learning_test.cpp) 'bin/mlpack_test
>> -t QLearningTest' should help to understand the overall structure. Also you
>> might find Shangtong's blog posts helpful:
>> http://www.mlpack.org/gsocblog/ShangtongZhangPage.html <http://www.mlpack.org/gsocblog/ShangtongZhangPage.html>
>> 
>> If you like you can work on a simple RL method like (stochastic) Policy
>> Gradients and use that to jump into the codebase, but don't feel obligated.
>> 
>> > I am thinking of working on my application at the earliest this week. Is that ok
>> > ? I am going through the code base and as I find something to talk about/on, can
>> > I trouble you people with my questions? There might be a lot, some even stupid !
>> 
>> Sounds like a good plan, let us know if we should clarify anything we are here
>> to help.
>> 
>> Thanks,
>> Marcus
>> 
>> > On 13. Feb 2018, at 19:08, Akash Shivram <akashshivram9 at gmail.com <mailto:akashshivram9 at gmail.com>> wrote:
>> >
>> > Hey there!
>> > Congratulations on getting into GSoC' 18!!
>> >
>> > I was going through the organisations participating this year searching for organisations working in ML and DL related field. I came across mlpack and was delighted to see a project on RL!! I like RL and and wanted some project to do in this field.
>> > I have experience working with Neural Networks, Reinforcement Leaning, and Deep Q Learning. As this is the first day of me with your repository,
>> > I have gone through requirements for an applicant for 'Reinforcement Learning' project and trying to go through as many papers listed as possible.
>> > Are there any more 'bonus' papers, or anything extra that wold be required.
>> > Moreover, I am thinking of working on my application at the earliest this week. Is that ok ? I am going through the code base and as I find something to talk about/on, can I trouble you people with my questions? There might be a lot, some even stupid !
>> >
>> > Thank you
>> >
>> > PS : This mail went too long!! Sorry for the long read !
>> > _______________________________________________
>> > mlpack mailing list
>> > mlpack at lists.mlpack.org <mailto:mlpack at lists.mlpack.org>
>> > http://knife.lugatgt.org/cgi-bin/mailman/listinfo/mlpack <http://knife.lugatgt.org/cgi-bin/mailman/listinfo/mlpack>
>> 
> 
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://knife.lugatgt.org/pipermail/mlpack/attachments/20180303/9a589125/attachment.html>


More information about the mlpack mailing list