ChanServ changed the topic of #mlpack to: "mlpack: a fast, flexible machine learning library :: We don't always respond instantly, but we will respond; please be patient :: Logs at http://www.mlpack.org/irc/
prudhvi-hack has joined #mlpack
prudhvi-hack has quit [Remote host closed the connection]
prudhvi-hack has quit [Remote host closed the connection]
prudhvi-hack has joined #mlpack
< prudhvi-hack>
Hello ,this is may be a stupid doubt .
< prudhvi-hack>
i used #define many times like #define x 10 like but why here we used #define without giving a value like #ifndef MLPACK_METHODS_CF_CF_HPP#define MLPACK_METHODS_CF_CF_HPP .I know that there will be a purpose of doing it but i can't found it by google search. please provide me a source explaining why it is needed
< prudhvi-hack>
sorry *not why it is needed but what it is doing ...
< zoq>
prudhvi-hack: Search for header guard.
< prudhvi-hack>
ok ,i will thank you
< prudhvi-hack>
yaa got it...
< zoq>
great
< taapas1128[m]>
<TaapasAgrawalGit "@zoq Hello, I am Taapas Agrawal "> zoq (@freenode_zoq:matrix.org): I am interested in the project on Reinforcement Learning. I am currently going through src/mlpack/methods/reinforcement_learning/. Is there any issue that you would recommend that I should solve that would guide me in the direction of the project ?
< kartikdutt18Gitt>
Hi @zoq, Sorry to disturb you again but if you get a chance could you tell me if I'm heading in the right direction with #2240 . Thanks.
< Nakul[m]>
I don't find currently issue specific for that.
< Nakul[m]>
But it good to work some easy issue to get through codebase and work flow
prudhvi-hack has quit [Remote host closed the connection]
prudhvi-hack has joined #mlpack
volhard[m] has left #mlpack []
< taapas1128[m]>
knakul853[m] : Thanks! I will find some issues to work on.
prudhvi-hack has quit [Ping timeout: 260 seconds]
togo has joined #mlpack
Mindjolt65 has joined #mlpack
Mindjolt has joined #mlpack
Mindjolt65 has quit [Remote host closed the connection]
< zoq>
I will take a look at the PR's and issues once I have a chance, no need to ping me here. We get a notification for very update; sometimes it takes some time before someone is able to provide feedback.
Ritik has joined #mlpack
togo has quit [Quit: Leaving]
Mindjolt has quit [Ping timeout: 260 seconds]
ImQ009 has joined #mlpack
Ritik has quit [Remote host closed the connection]
< rcurtin>
agreed with zoq; sorry that we can't review everything immediately, but we'll get to it as we're able to
< rcurtin>
in this community we have far more contributors than maintainers, so there is a bottleneck; unfortunately not too much we can do about it (I guess I could quit my day job but that seems unwise :-D)
prudhvi-hack has joined #mlpack
< metahost>
Hi, Marcus, Ryan, how should I go about with a proposal? Will I share a link to the mailing list for review and suggestions?
< rcurtin>
metahost: sure, that works; have you seen the proposal guide on Github? that should be helpful too
< rcurtin>
typically what I look through a proposal for is (a) clear description of the work to be done; (b) clear description of what the API will be like (we have to be sure it fits will with the rest of mlpack); (c) clear and realistic timeline
< rcurtin>
for (c) I should be careful writing that, as a "realistic timeline" is basically impossible to come up with :)
< rcurtin>
so the bounds of "realistic" are pretty large---but basically what I mean is, if it says "I'll reimplement the entire codebase of mlpack in Scala in one week", well, that's a bit unrealistic ...
< jeffin143[m]>
Haha
< jeffin143[m]>
The toughest part in writing a code is naming variable
< jeffin143[m]>
Also anyone till now fought with CORS errors ???
< jeffin143[m]>
It's more difficult than a machine learning algorithm
< rcurtin>
I do everything I can to avoid web development for exactly that reason :-D
< rcurtin>
ML is easy, web is hard
< PrinceGuptaGitte>
@rcurtin since you're explaining things about GSoC, I had a doubt. How large should the project be? For example I thought of implementing YOLO algorithm for object detection
< PrinceGuptaGitte>
But I'm confused if it's enough for 3 months of work
< PrinceGuptaGitte>
It'll have a lot of things to do like non max suppression, IOU, implementing anchor boxes, but is that a good enough project
Mindjolt has joined #mlpack
< zoq[m]>
Absolutely a big enough project.
< PrinceGuptaGitte>
ML is also more fun, in web we keep using APIs all the time. Not that fun.
< metahost>
rcurtin: yes, I went through the application guide! :)
< metahost>
Also, thanks for the detailed explanation!
< rcurtin>
:) sure, happy to help
< rcurtin>
and I agree with marcus, that size of project seems to be perfectly large enough to me too
< rcurtin>
one thing to keep in mind is that timelines are often over-optimistic :) so typically I expect the students to not get 100% to the end of their project :)
< rcurtin>
for that reason, I'm often looking to see some 'partial' goals, so that even if the project isn't completed, *something* useful is still merged into the codebase
< rcurtin>
of course, that's not always possible, but when it is possible to plan for various checkpoints that can be merged into master before the end of the summer, that generally seems to lead to more successful projects
< SriramSKGitter[m>
In that case, is a project implementing multiple smaller algorithms more favourable than a single large one?
< PranavReddyP16Gi>
Hey everyone I was thinking of doing something related to NEAT or other such NeuroEvolution algorithms. Can someone give me a bit of guidance regarding that? Also is it possible to make a project suitable for gsoc related to NE? Originally I didn't think I would be trying for gsoc but due to a change in circumstaces I thought I'd give it a go.
< rcurtin>
SriramSKGitter[m: not necessarily, I didn't mean to imply that---there are some projects for which intermediate merges might not be reasonable and that's okay :)
< rcurtin>
all I mean is that if it *is* possible to arrange the project in a way that has intermediate 'checkpoints', that's usually a good idea to do
< rcurtin>
but remember, we are humans looking at the applications and so we can appreciate when there's nuance to the situation :)
< jeffin143[m]>
I am a great example of that :)
< jeffin143[m]>
It's almost a year we are yet to merge my gsoc work :)
< jeffin143[m]>
Hopefully it will get merged soon
< rcurtin>
jeffin143[m]: I saw Mikhail's updates, really excited about it
< rcurtin>
hoping to merge soon... we are almost there :)
< jeffin143[m]>
:)
akashd has joined #mlpack
prudhvi-hack has quit [Remote host closed the connection]
< PranavReddyP16Gi>
@zoq https://github.com/mlpack/mlpack/pull/1908 applies NEAT right? How about implementing other neuroevolution algorithms like CNE or HyperNEAT? I don't know too much about it yet but I'm really starting to get into the whole thing.
akashd has quit [Quit: akashd]
akashd has joined #mlpack
< ShikharJaiswalGi>
@rcurtin Ryan, are you there
< ShikharJaiswalGi>
(edited) ... you there => ... you there?
akashd has quit [Client Quit]
< zoq>
PranavReddyP16: We already have a CNE implementation in ensmallen, but adding another neuroevolution method sounds good, HyperNEAT might be an option.
< rcurtin>
ShikharJaiswalGi: yep, I am here right now, just got back from lunch :)
< Nakul[m]>
Hey abernauer
< Nakul[m]>
are you there ?
< PranavReddyP16Gi>
@zoq got it. As of now I understand that the key difference between NEAT and HyperNEAT is that HyperNEAT involves an indirect encoding scheme as compared to NEAT's Direct encoding. I will read about it in a lot more detail. Until now all my attention has been directed towards NEAT. I didn't even know other NE algorithms existed until very recently. I will delve more into the topic
< zoq>
Sounds good :)
godfather_rana has joined #mlpack
< godfather_rana>
hey I am new here, can anyone guide me how to contribute
< ShikharJaiswalGi>
I might be able to make a decision sometime next week.
< rcurtin>
ShikharJaiswalGi: there's no hurry at all, the final decision is whether or not to accept a student :) so that would be mid-to-late April I think
Mindjolt has quit [Ping timeout: 260 seconds]
travis-ci has joined #mlpack
< travis-ci>
mlpack/ensmallen#693 (master - 08994c5 : Marcus Edel): The build has errored.
< ShikharJaiswalGi>
@rcurtin Cool, I'll let you know as soon as I sort my summer plans out.
< vigsterkr[m]>
/msg rcurtin so how many ppl are nagging u? :)
< vigsterkr[m]>
mmm not irc :D
< vigsterkr[m]>
/rcurtin
< vigsterkr[m]>
mmm this matrix...
< vigsterkr[m]>
:)))
< kartikdutt18Gitt>
Ohh, really sorry about that. Thanks.
< PrinceGuptaGitte>
I think namespaces helps to avoid them, unless cors isn't what I think.
< rcurtin>
hahaha :-D
< rcurtin>
ShikharJaiswalGi: no worries, that sounds good. co-mentoring is always a possibility too, but anyway, up to you :)
< PranavReddyP16Gi>
Also @zoq just a curiousity, Will implementing hyperneat be sufficient as a gsoc project or should I approach this as a stand-alone type of implementation
TarunLahrodGitte has left #mlpack []
ImQ009 has quit [Quit: Leaving]
< zoq>
PranavReddyP16Gi: This could work as a GSoC project, yes.
< vigsterkr[m]>
@zoq keep up the stream! :)
< vigsterkr[m]>
currenly on #shogun i'm the only one who's handling the flood :)))
< zoq>
vigsterkr[m]: technically, I'm in that channel as well :)
< vigsterkr[m]>
help! :D
< vigsterkr[m]>
hahahah
sipun has joined #mlpack
< SaraanshTandonGi>
@zoq Could you please clarify #2236 requested change?
Pranshu54 has joined #mlpack
Pranshu54 has quit [Remote host closed the connection]