verne.freenode.net changed the topic of #mlpack to: http://www.mlpack.org/ -- We don't respond instantly... but we will respond. Give it a few minutes. Or hours. -- Channel logs: http://www.mlpack.org/irc/
Bartek has quit [Remote host closed the connection]
Neuron1k has quit [Ping timeout: 250 seconds]
toshad has quit [Quit: Connection closed for inactivity]
vladx has joined #mlpack
vladx has quit [Client Quit]
< decltypeme>
rcurtin how do we get confidence scores of the Naive Bayes Classifeir for example?
< rcurtin>
unfortunately that support doesn't exist... it should definitely be added, I just have not had the time to do that
< rcurtin>
if you wanted to add that support I would definitely accept the patch... :)
< decltypeme>
yes, i may work on it today actually, will let you know
Pooja has joined #mlpack
< rcurtin>
decltypeme: sounds good, if you get it working, open a PR :)
< rcurtin>
I think that it should have the same API as the Hoeffding tree, which is the only other algorithm that supports confidence scores
< rcurtin>
but I don't know if the Hoeffding tree is the right way to do it...
< rcurtin>
sorry, the function I'm referencing is in my own personal fork
< rcurtin>
anyway, maybe the right way to do this is
Pooja has quit [Ping timeout: 250 seconds]
< rcurtin>
Classify(MatType& points, arma::Row<size_t>& predictions /* fill this with class predictions */, arma::rowvec& probabilities /* confidence estimates of most probable class */);
< rcurtin>
and maybe also
< rcurtin>
Classify(MatType& points, arma::Row<size_t>& predictions, arma::mat& probabilities /* confidence scores for each class for each point */);
< rcurtin>
I dunno, if you think of something better, go ahead and do that :)
< decltypeme>
I was thinking actually of breaking down Classify() into Scores() and letting Classify() call Scores() ... because I wanted to maintain the Classify's signature for backward compatiblities but if you think it won't be a problem it would be better to wrap it in one function
< rcurtin>
oh, I was thinking, just provide some extra overloads of Classify()
< decltypeme>
yes, seems great thank you :)
thesanefreak has joined #mlpack
< decltypeme>
rcurtin: I shared a proposal draft, Still will add a preamable, about me and elaborate more about the project. however if you have initial thoughts or suggestions, please share :)
thesanefreak has quit [Quit: Page closed]
thesanefreak has joined #mlpack
wasiq has joined #mlpack
< thesanefreak>
Hi, I was going through the idea list on github and wanted to know if "Implement tree types" is still open?
Pooja_ has joined #mlpack
< Pooja_>
hello
Pooja_ has quit [Client Quit]
thesanefreak has quit [Quit: Page closed]
agobin has quit [Quit: Connection closed for inactivity]
nataliat95 has quit [Ping timeout: 244 seconds]
ftuesca has quit [Quit: Leaving]
miku_ has joined #mlpack
Mathnerd314 has quit [Ping timeout: 268 seconds]
wajibat has joined #mlpack
wajibat has quit [Ping timeout: 250 seconds]
< miku_>
rcurtin: Gentle reminder, please review my application when you have the time :)
Jeffu_u has joined #mlpack
nilay has joined #mlpack
< nilay>
what is meant by 1x1 convolution or 5x5 convolution?
< nilay>
zoq any help would be appreciated.
nilay_ has joined #mlpack
Jeffu_u has quit []
nilay has quit [Ping timeout: 250 seconds]
nilay_ has quit [Quit: Page closed]
wasiq has quit [Ping timeout: 244 seconds]
nilay has joined #mlpack
miku_ has quit [Ping timeout: 250 seconds]
miku_ has joined #mlpack
satyaki3794 has joined #mlpack
anatolemoreau4 has joined #mlpack
wasiq has joined #mlpack
nilay has quit [Ping timeout: 250 seconds]
agobin has joined #mlpack
anatolemoreau has joined #mlpack
anatolemoreau4 has quit [Ping timeout: 264 seconds]
toshad has joined #mlpack
AtlzKb has joined #mlpack
arnab has joined #mlpack
nilay has joined #mlpack
kienhuynh has joined #mlpack
kienhuynh has quit [Client Quit]
< nilay>
Hello zoq, I have a question, are you there?
Keon has quit []
satyaki3794 has left #mlpack []
agobin has quit [Quit: Connection closed for inactivity]
anatolemoreau1 has joined #mlpack
anatolemoreau has quit [Ping timeout: 268 seconds]
Nilabhra has joined #mlpack
< rcurtin>
miku_: I'll take a look when I get to work in about 45 minutes
< nilay>
can anyone guide me on how to describe the API of the project in the proposal, is there a set format, or should I just write approximately the functions that are required.
< nilay>
I am sorry if this is a bit late, but any help is much appreciated.
Bartek1 has joined #mlpack
arnab has quit [Quit: Page closed]
< zoq>
nilay: Hello, take a look at: https://github.com/mlpack/mlpack/wiki/DesignGuidelines, a detailed description of the overall API isn't necessary. Although it's a good idea, if you describe the important parts of the API clearly.
< miku_>
rcurtin: great, thanks :) In particular, I was a little unsure about how to technically detail the API and Testing sections, so if you could look at that first, it'd be great.
anatolemoreau2 has joined #mlpack
anatolemoreau1 has quit [Ping timeout: 268 seconds]
< wasiq>
rcurtin, hi in the test_functions.hpp we use 2 test fuctions that were taken from 2 separate papers. what kind of functions do you plan on implementing for the FW algorithm? There is a paper on low rank solutions
< rcurtin>
wasiq: the test functions chosen for the FW algorithm are up to the student implementing them; I personally have no preference
< rcurtin>
miku_: ok, I left some comments (I think you saw); I won't have time for any more, so hopefully what I've written is helpful
< rcurtin>
keep in mind that we can get in touch with you to ask for questions or clarifications after the proposal deadline
< miku_>
Sounds good, thanks
cheral has joined #mlpack
< Nilabhra>
rcurtin: can the proposal be improved after the deadline? I mean by discussion with mentor?
< nilay>
zoq: any advice on how to choose ( or papers to look for) the test functions for we need to go deeper: GoogLeNet project. Testing backpropagation in the intermediate layers having auxillary classifiers
< rcurtin>
Nilabhra: if necessary, yes, but the mentors will get in touch with you about that
< Nilabhra>
rcurtin: great! I assume this is before the proposals are submitted to google?
< rcurtin>
what do you mean?
< rcurtin>
when you submit the proposals on the summer of code website, you are submitting them to the mentoring organization
< rcurtin>
google never actually looks at the proposals; that's up to the organization
< rcurtin>
decltypeme: I like the idea of ensemble methods, but make sure to clarify exactly which ensemble methods you are planning on implementing
< rcurtin>
mlpack already has AdaBoost in src/mlpack/methods/adaboost/, so it may be worth looking at that
< rcurtin>
decltypeme: also be sure that you submit your final proposal as a PDF, not as a link to a github wiki
< rcurtin>
Nilabhra: sorry, I was incorrect. last year, changing the proposal was possible, but this year, Google is not allowing it
< Nilabhra>
rcurtin: Oh, well I will hope for the best then.
< rcurtin>
I think it's still reasonable for mentors to send you any emails or anything to ask for clarifications, but I'll have to check that Google allows that
< rcurtin>
Nilabhra: my only comment is, can you elaborate how the FM code will work with the CF code? one of the more important things for us when reviewing these proposals is how easily what students are making will fit with what we already have
< Nilabhra>
rcurtin: ok sure, I'll elaborate that :) thanks a lot
rosy_ has joined #mlpack
< rosy_>
GSOC mentor? anybody?
ftuesca has joined #mlpack
< zoq>
rosy_: Hello, yes
anatolemoreau3 has joined #mlpack
anatolemoreau2 has quit [Ping timeout: 248 seconds]
Mathnerd314 has joined #mlpack
rosy_ has quit [Ping timeout: 252 seconds]
anatolemoreau3 has quit [Ping timeout: 268 seconds]
< decltypeme>
rucrtin: thank you for getting back, i will elaborate more on the certain types of ensembles to be implemented, but is the API representation clear or you need me to elaborate more?
rosy__ has joined #mlpack
< rcurtin>
decltypeme: it seems clear enough to me. I wonder if maybe the fusion method should be a template parameter for the ensemble class, instead of just a single parameter to the Train() method
< rcurtin>
but of course that is something that can be discussed and changed later if necessary
< decltypeme>
yes, that maybe another way of doing it, i will note it out there
nilay has quit [Ping timeout: 250 seconds]
qwebirc15291 has joined #mlpack
< qwebirc15291>
Is this channel for helping people with the GSoC project?
< decltypeme>
yes, and other mlpack things :)
< qwebirc15291>
Okay. I have a doubt which I am kind of embarrassed to ask. But still I am going to go ahead and ask it.
palashahuja has joined #mlpack
< palashahuja>
zoq, hi
skon has joined #mlpack
< qwebirc15291>
In the aggregate stored function project we have to write a C code for allowing the users to create new aggregate functions right?
< rcurtin>
qwbirc15291: I'm not sure which project you are talking about
< qwebirc15291>
Ummm.. Okay there is there project mentioned by MariaDB which is for creating Aggregate Stored Functions.
< rcurtin>
I think maybe you are in the wrong channel...
< rcurtin>
this is the mlpack channel not MariaDB :)
< qwebirc15291>
Oh! Sorry :P
< rcurtin>
no worries :)
< qwebirc15291>
Well since I am here.. I am taking a course on data mining this year and know C++. But I have no clue about machine learning. Any suggestions regarding which project I should apply for?
< skon>
rcurtin , are you there ?
Awcrr has joined #mlpack
qwebirc15291 has quit [Quit: Page closed]
rosy__ has quit [Ping timeout: 264 seconds]
skon has quit [Ping timeout: 244 seconds]
< zoq>
qwebirc1529: Choose the project you're most passionate about. If you apply for one of the mlpack projects, the "necessary knowledge" sections can often be replaced with "willing to learn", so you don't have to be an machine learning expert.
< zoq>
palashahuja: Hello
< palashahuja>
Could you please explain as to what else are you looking out for in the basic information ?
skon has joined #mlpack
< skon>
rcurtin , are you there ?
agobin has joined #mlpack
< zoq>
palashahuja: You are fine; If you like you could answer some of the optional questions, but don't feel obligated.
< skon>
hello zoq
nilay has joined #mlpack
< zoq>
skon: Hello, you can always ask your question here and we get back to you once we have the time.
< skon>
sry , my internet has been fluctuating a lot recently , so i was not quite sure if i was connected
< zoq>
skon: no worries
skon has quit [Client Quit]
anatolemoreau3 has joined #mlpack
Awcrr has quit [Ping timeout: 252 seconds]
< palashahuja>
zoq, if you could evaluate the proposal one last time that would be great
< palashahuja>
would be really grateful if you could do so.
< miku_>
rcurtin: Will be great if you'd look at the redesigned API before I submit.
cheral has quit [Quit: Connection closed for inactivity]
< zoq>
palashahuja: I'm trying to do that, keep in mind that we can get in touch with you to ask for questions or clarifications after the proposal deadline.
< palashahuja>
zoq , ok cool
anatolemoreau3 has quit [Ping timeout: 264 seconds]
< virtualgod>
Hello everyone. I am Varun Bhatt, a second year undergraduate student in Indian Institute of Technology, Bombay, India. Sorry for being late but I have uploaded my proposal draft on GSoC website for the project on neuroevolution algorithms. I saw the discussions about this project in the mailing list archives and hopefully I have included all the required
< virtualgod>
things in the proposal. Kindly review my proposal.
rosy_ has joined #mlpack
< palashahuja>
thanks for your time zoq
< palashahuja>
I will go ahead and submit it :)
< zoq>
palashahuja: sounds good
< palashahuja>
Sorry for the annoying permissions issue :?
< palashahuja>
:)
AtlzKb has quit [Ping timeout: 250 seconds]
anatolemoreau3 has joined #mlpack
rosy_ has quit [Ping timeout: 260 seconds]
< nilay>
zoq: can you please review my proposal before I submit it.
< nilay>
Thank You to all the mentors for their support up until now. You guys are doing a great job. :)
< palashahuja>
I agree with nilay on this
ank_95_ has joined #mlpack
< awhitesong>
zoq, rcurtin: Hey! so i have uploaded my final proposal in essential deep learning modules. (pdf and draft). Could you please review it. Thanks!
agobin has quit []
agobin has joined #mlpack
nilay has quit [Ping timeout: 250 seconds]
AtlzKb has joined #mlpack
anatolemoreau3 has quit [Ping timeout: 260 seconds]
< palashahuja>
has anyone watched the new batman vs superman movie ?
< palashahuja>
looking for reviews here
mentekid has joined #mlpack
AtlzKb has quit [Ping timeout: 250 seconds]
< rcurtin>
palashahuja: I didn't see it, but I heard that it wasn't good from some friends
< rcurtin>
I have no idea if their opinion is accurate :)
< palashahuja>
hmm
< palashahuja>
but I have also heard that the ending was great ..
< rcurtin>
maybe, I am clueless :)
< palashahuja>
same here :)
rebeka has joined #mlpack
< awhitesong>
30% on rottentomatoes though
< palashahuja>
hmm
miku_ has quit [Ping timeout: 250 seconds]
Neuron1k has joined #mlpack
kirizaki has joined #mlpack
< kirizaki>
Hi
< rcurtin>
hello!
< wasiq>
Its good.
< wasiq>
Its a bit vague in the 1st half but the 2nd half is worth it.
< wasiq>
I haven’t seen it either though. Friends opinions.
rosy has joined #mlpack
< palashahuja>
thanks for the opinion wasiq
skon has joined #mlpack
< skon>
zoq, rcurtin , I know its late can you guys have a look at my proposal.
< skon>
?
< rcurtin>
with only 15 minutes left I doubt I will be able to provide much useful input... can you send a link to the document?
< skon>
i have some problems with my isp ,luckily was able to submit
< rcurtin>
yeah, I am glad you were able to submit
< skon>
changing isp tomorrow
Bartek1 has quit [Ping timeout: 248 seconds]
< rcurtin>
it looks good, it looks like you detailed everything including your timeline
< rcurtin>
hopefully the new ISP will be better :)
< skon>
yes
decltypeme has left #mlpack []
decltypeme has joined #mlpack
< decltypeme>
thank you rcurtin and the dear mentors
< decltypeme>
it has just *started*
< skon>
yes , thanks a lot for the support and answering stupid questions :p
sumedhghaisas has joined #mlpack
< rcurtin>
thanks again students for submitting proposals
awhitesong has quit [Remote host closed the connection]
< rcurtin>
it looks like we have received 119 this year, which is just about twice as many as we've ever had...
awhitesong has joined #mlpack
tsathoggua has joined #mlpack
< wasiq>
Good luck everyone. Thank you mentors for being so patient :)
< rcurtin>
no problem, it is always fun (and challenging because of time) to meet so many talented people :)
miku_ has joined #mlpack
< rebeka>
Thanks mentors for all the help. I have a question though :P For those of us who didn't get time for our proposals to be reviewed can we expect a review now or do we have to wait till 22 April for the final verdict?
< rcurtin>
rebeka: the mentors can get in touch if they have any questions or need any clarifications, otherwise, yeah, you'll hear back on April 22 (or whenever google sends out the notifications)
anatolemoreau3 has joined #mlpack
< rebeka>
thanks :)
rebeka has quit [Quit: Page closed]
< awhitesong>
rcurtin: hey. were you able to see the proposal i submitted ?