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/
Mathnerd314 has quit [Ping timeout: 268 seconds]
Bang has joined #mlpack
Bang is now known as Guest64891
Guest64891 has quit [Client Quit]
mentekid has joined #mlpack
Nilabhra has joined #mlpack
Nilabhra has quit [Remote host closed the connection]
< mentekid>
hey, did anyone upgrade to Ubuntu 16.04 and have problems with armadillo?
< zoq>
mentekid: hmm, can you tell us more about your problems with armadillo?
< mentekid>
I solved it by linking libarmadillo.so.6 to libarmadillo.so.5, which had disappeared
< zoq>
ah, okay
< mentekid>
because when I tried to reinstall libarmadillo5 it had no install candidate
< mentekid>
but it was weird, libarmadillo6 was there but mlpack was just looking for version 5
< mentekid>
and it was working before the upgrade, that's why I asked if someone shared the problem
< zoq>
Did you remove the old CMake files before rebuilding mlpack?
< mentekid>
Do you mean rebuilding after the upgrade? I didn't, should I have?
< zoq>
You configured mlpack with the old armadillo version which is in a specific location, but as you said you updated ubuntu/armadillo, so I think the problem is mlpack searches for that old armadillo lib and can find it. So I think rerunning "cmake .." should find the updated version and everything is fine.
< zoq>
Linking manually against the old lib also works.
< mentekid>
ah i see
< mentekid>
no I mean I created a symlink of v6 and naming it what mlpack was looking for (v5). It's a bit hacky though so I'll do what you said instead
< mentekid>
named it*
< mentekid>
thanks :)
< zoq>
Sure, I guess I see you here in a couple of hours :)
< zoq>
marcosirc: since this is a logged channel, you should probably change your password
< marcosirc>
Hi, yes. I added a space to the beginning of last command!
nilay_ has joined #mlpack
< rcurtin>
oops! :(
Guest42582 has quit [Ping timeout: 250 seconds]
nilay has quit [Ping timeout: 250 seconds]
nilay_ has quit [Ping timeout: 250 seconds]
nilay has joined #mlpack
nilay_ has joined #mlpack
nilay has quit [Ping timeout: 250 seconds]
Bang has joined #mlpack
tham has joined #mlpack
sumedhghaisas has joined #mlpack
sumedhghaisas has quit [Client Quit]
sumedhghaisas has joined #mlpack
nilay_ has quit [Ping timeout: 250 seconds]
nilay has joined #mlpack
Bang is now known as Guest28091
nilay_ has joined #mlpack
< rcurtin>
okay, I guess it's 1500 UTC and everyone is here, so we should get started :)
< rcurtin>
so, first of all, congratulations to the students on your acceptance to GSoC!
< rcurtin>
this was the most selective year we've had for GSoC, with 119 applications and only 6 slots
< rcurtin>
maybe you have seen it already, but if you are interested in the details of each of the six projects, I made a short writeup in an email to the mailing list:
< rcurtin>
so, currently we are in the "Community Bonding" period, which goes from April 22 to May 22
< rcurtin>
then, the actual coding is from May 23 to August 23, and at some point there is a "midterm" which basically is just a check that things are going relatively according to plan (more on this later)
< rcurtin>
after that, the results are announced on August 30, and that concludes the summer
< rcurtin>
if there are any stipend issues or anything like this, Google would be who to talk to, but you can get in touch with me if you like and I can try to help out :)
< rcurtin>
after the summer is over, we hope that everyone will stick around and be a part of the mlpack community, and maybe be a student again next year, or a mentor
< rcurtin>
so, moving on, I think it's important to lay down expectations for students and mentors
< rcurtin>
before I do that, I want to point out that I don't think these expectations will be too relevant, since we think you are all great students, but it's important to set groundwork for expectations, in case it needs to be revisited later
< rcurtin>
and, if at any time you have a question, feel free to interrupt me :)
< rcurtin>
so, student expectations --
< rcurtin>
students are expected to be doing the equivalent of a full-time job, so, a full work week
< rcurtin>
it's okay if some weeks you work more or less than others, but in the end the amount of time put in should be roughly equivalent to a full-time job
< rcurtin>
regular contact and communication with mentors is important and expected; preferably via IRC, but email is okay too (this is up to the student and mentor; use whatever is best)
< rcurtin>
IRC is real-time, so it's usually better for debugging. usually, there are at least a few people (including zoq and myself) in #mlpack who are here to help out
< rcurtin>
also, there's no problem reaching out for help to someone other than your mentor. we are a community, after all :)
< rcurtin>
students are also expected to provide some kind of weekly status update to the community
< rcurtin>
this could be an email post, like this one from Marcus in 2013:
< rcurtin>
whichever format you use is up to you, but weekly updates are important because more folks in the community than just your mentor may want to be reading about how the project is going
< rcurtin>
if you do want to do blog posts, we used a Github repository for that last year, at https://github.com/zoq/blog
< rcurtin>
next, if you are going on vacation or will be unavailable for some period of time, please give advance warning to your mentors
< rcurtin>
disappearing students is a big issue with GSoC projects, so we would like to avoid that :)
< rcurtin>
at the same time, there is no need to tell your mentor every time you get up to go get lunch or anything; but, you should at least be in regular contact with your mentor
< rcurtin>
not every project goes according to the proposal; sometimes, the reality of the project is more difficult than what was anticipated
< rcurtin>
this is okay! if this happens, the student and the mentor should work together to reassess what is feasible and adjust the plan accordingly
< rcurtin>
we really hope we don't have to fail any students this year; but, we will fail a student who disappears, doesn't appear to be working the expected amount of time, or who is otherwise seriously underperforming
< rcurtin>
if any of those situations occur, the student will be made aware with warnings, so failure will not be unexpected
< rcurtin>
personally, I don't expected that this will be an issue, but, I think it's really important to make all of this clear before we get started
< rcurtin>
any questions before I move on? I know I wrote a lot :)
< zoq>
no, sounds good to me :)
< mentekid>
clear to me too
< tham>
sounds good to me too :)
< rcurtin>
okay, well I will move on to mentor expectations then
< nilay_>
1
< keonkim>
cool!
< Guest28091>
Good to me too
< marcosirc>
Ok for me too.
< rcurtin>
just like students should be in regular contact with their mentors, mentors should also be in regular contact with their students, and able to provide reasonably quick responses
< rcurtin>
(i.e. within a day or two, if not in real-time)
< lozhnikov>
Clear
< tham>
ok
< rcurtin>
mentors should work out with their students times in which they are both available regularly to work with each other, and they should be willing to help debug technical and code problems
< rcurtin>
the mentor shouldn't be doing the majority of the work, of course, but mentors should be there to help when needed
< rcurtin>
mentors should also give their students advance warning if they will be out of touch for a while
< rcurtin>
this is relevant to me, because I am often out of town on the weekends :)
< rcurtin>
I will keep students who I am mentoring aware of my schedule so it doesn't seem like I've just disappeared
< rcurtin>
preferably, when possible, GSoC communication should be done in public (on Github, IRC, mailing list, whatever is fine); it can often be better to ask a question publicly, because then many people can answer
< rcurtin>
of course, that isn't applicable in every case
< rcurtin>
that does it for the mentor expectations; any questions before I move onto a little bit of mlpack history and other miscellany? :)
< sumedhghaisas>
aye aye captain :)
< rcurtin>
:)
< rcurtin>
okay, some quick history...
cache-nez has joined #mlpack
< rcurtin>
mlpack is not a new project; it was originally developed in 2007/2008 by a lab at Georgia Tech called "The FASTLab"
< rcurtin>
I think the website is still up... fast-lab.org ?
< rcurtin>
yeah, it's still up. the website is pretty horrible and hard to read...
< rcurtin>
anyway, the website is the same way the code was -- it was just a disorganized collection of research algorithms
< rcurtin>
when I joined the lab in late 2009 / early 2010, I was given the task of maintaining the library
< rcurtin>
so during most of 2010 and 2011 I worked with a team of people (mainly James Cline and Neil Slagle, who aren't active contributors anymore) reworking basic design decisions
< rcurtin>
this culminated with a design document, which I think I sent out at some point, but if you are interested, it is at http://www.igglybob.com/mlpack_future.pdf
< rcurtin>
I think maybe that is kind of out of date now, and more recently we have redone some of the design guidelines on Github:
< rcurtin>
anyway, at the end of 2011, mlpack was presented at a NIPS workshop, and it was published in JMLR in 2013
< rcurtin>
more importantly, in 2013, mlpack was accepted into GSoC
< rcurtin>
we were also a part of GSoC in 2014 too
< rcurtin>
so from about 2013 onwards, mlpack development has shifted from being a small group at Georgia Tech to a large worldwide community with users on every continent
< rcurtin>
(except Antarctica, but it's possible someone is using mlpack there and simply hasn't gotten in touch, so we don't know)
< rcurtin>
and now we have 50+ contributors
< rcurtin>
when I talk to researchers at conferences, more and more are familiar with mlpack, and comments I hear about mlpack are always very good
< rcurtin>
so, having watched mlpack grow from a small project to a fairly well-known library has been very exciting, and I'm hopeful that GSoC 2016 will continue to add to this momentum
< rcurtin>
and I'm happy that we can all be a part of this :)
< rcurtin>
in addition, I graduated from Georgia Tech (finally) and now work at Symantec, and Symantec is willing to pour some good amount of resources into mlpack, so we have good support
< rcurtin>
this was used with Jenkins to build mlpack with a wide variety of different build configurations (like different versions of Armadillo, different architectures, etc.)
< rcurtin>
but when I left Georgia Tech, this had to come down. I still have a few systems up, hosted at Symantec: http://big.mlpack.org:7780/
< tham>
wow,never though it need so many machine
< rcurtin>
in the next few weeks, a system called "masterblaster" will come online; it has 72 cores, I think 1TB RAM, and expensive SSDs
< rcurtin>
tham: yeah, we would build on something like 60 different configurations a night, maybe more
< rcurtin>
lots of different versions of Armadillo
< keonkim>
:O
< rcurtin>
and all those computers were kind of slow and old, so many were necessary :)
< rcurtin>
anyway, masterblaster will take the place of basically all those systems, but it may be a little while until it is up, because there is a lot of bureaucracy to contend with at Symantec
< rcurtin>
I also have some other build slaves I am trying to set up; I found three Sun Enterprise ULTRASparc T5220s in a closet here, and when I have a chance I work on configuring them to bring them online. those will be another 64 build cores each
< rcurtin>
due to the state of the build systems, these benchmarks have not been run in a while. But I'm hoping to put sometime into them this summer
< rcurtin>
some of the Shogun toolbox GSoC students will also be working with this system too
< rcurtin>
if you're interested in build servers and things like this, there is definitely room for help, so let me know if you want to help out getting these set up :)
< rcurtin>
next, I'm hoping to release mlpack 2.0.2 in the next couple weeks. I've been wanting to have more frequent releases, but it's been difficult to find the time
< rcurtin>
if anyone is interested in helping out with the release process, let me know. I don't think it's very exciting though...
< rcurtin>
okay, that's everything I have in my notes. any questions? anything I forgot to talk about? :)
< rcurtin>
(then we can introduce ourselves, but I wanted to save that bit for last)
< mentekid>
is 1TB of RAM enough to run Google Chrome? Or do you still get lag? :P
< rcurtin>
haha
< mentekid>
jk no questions here
< Guest28091>
Are we going to use the super machine for building mlpack?
< rcurtin>
I run chrome on a 4K display here at work, it's pretty horrible, I can barely load Google maps... and this system has 128GB RAM... maybe I need to upgrade to 1TB!
< rcurtin>
Guest28091: yeah, the idea is that it will run Jenkins (like http://big.mlpack.org:7780), and each night it will build mlpack with different versions of armadillo, different compilers, different versions of boost, etc.
< rcurtin>
okay, I guess there are no more questions; if you think of anything later, feel free to ask here in #mlpack or send an email or whatever, anything works :)
< rcurtin>
so I think it's a good idea to introduce ourselves now so that we can get to know each other a little bit, since we will be working with each other all summer
< Guest28091>
Sorry I have a question
< rcurtin>
ah okay :)
< Guest28091>
Thanks! My concern is: my experience on compiling large open source project like mlpack is limited
< Guest28091>
Is it possible to offer some tutorial for using the building tools?
< rcurtin>
and if not, let us know, and we can update that tutorial or write a new one
< Guest28091>
Yeah, so that is it! I tried this and already successed
< rcurtin>
you don't need to test every version of armadillo on your work system
< Guest28091>
Are we all going to use one public machine for testing or just use our own laptop is OK?
< rcurtin>
that's what the build server is for --- when you check in code, travis and AppVeyor will make sure it builds (in the PR), and then every night Jenkins will check to make sure it builds on all configurations
< rcurtin>
and if it doesn't, we can figure out what the issue is and fix it from there
< Guest28091>
Sounds great!
< Guest28091>
No more questions for me currently. Thanks!
< rcurtin>
okay, great
< rcurtin>
so, introductions: I'm Ryan; I live in Atlanta, Georgia. Right now I work at Symantec as a machine learning researcher, and I'm also able to put a lot of my time towards mlpack development, which is really nice
< rcurtin>
In my free time, I like to do woodworking and drive old cars around the countryside and take cheesy pictures of old bridges: http://ratml.org/misc_img/saluda_bridge.jpg
< rcurtin>
(I'm the bottom left)
< rcurtin>
I guess I should clarify, I mean Georgia in the USA not Georgia the country :)
< rcurtin>
I didn't think of any particular formal process for introductions or anything, so, I guess, introduce yourself if you like :)
< sumedhghaisas>
thank you for clarifying that :)
< sumedhghaisas>
Okay I can introduce myself now...
< sumedhghaisas>
I am Sumedh Ghaisas...
< sumedhghaisas>
I am working in Ebay...
< sumedhghaisas>
I think everyone knows what that is. :)
< keonkim>
yup :)
nilay_ has quit [Ping timeout: 250 seconds]
< sumedhghaisas>
I joined MLPACK in around 2014 GSOC... maybe 5 or 6 months before that... I did a project in the same
nilay has joined #mlpack
< sumedhghaisas>
umm.. past time.. playing table tennis... listening to blues... and coding of course...
< sumedhghaisas>
that would be it :)
< sumedhghaisas>
and yes happy to be part of GSoc again :)
< rcurtin>
I like the blues, I need to find some more good blues artists to listen to :)
< mentekid>
So you're a mentor, right?
< sumedhghaisas>
hey.. I can compile a very good playlist for you...
< sumedhghaisas>
if you want??
< rcurtin>
sure I would like that :)
< sumedhghaisas>
yup... I will mentoring marcos...
< sumedhghaisas>
you got it :)
< Guest28091>
Hello guys! My name is Bang Liu, and I am doing the "Neuroevolution Algorithms Implementation" with my Mentor Marcus Edel. Great to meet all of you and excited about this interesting project and summer!
< Guest28091>
I am currently a first PhD student at University of Alberta, Canada. I am interested in researches related to machine learning. I have a homepage, www.ualberta.ca/~bang3. You can visit if interested.
< Guest28091>
When I am free, I like drawings (though not good...), readings and watching movies; I also like swimming, playing Pingpang, badminton.
< rcurtin>
maybe you and sumedh should play each other sometime :)
< Guest28091>
Haha, yeah
< sumedhghaisas>
ryan will sponsor that match :)
< rcurtin>
:)
< zoq>
Okay, my turn. Marcus, PhD student at the Free University of Berlin here in Germany. Working on different projects that are all kind of related to self-driving cars. In my free time I like to go paragliding, I guess I could upload some fancy images if anyone likes to see them :)
< rcurtin>
I would love to look at that, paragliding sounds really cool
< mentekid>
oooh I always wanted to do paragliding
nilay_ has joined #mlpack
< zoq>
you should do that, it's awesome :)
< tham>
It looks cool, but I am too afraid to do it
nilay has quit [Ping timeout: 250 seconds]
nilay_ has quit [Client Quit]
< keonkim>
ok everyone disappeared or is it my computer
< rcurtin>
it looks like nilay left, but there are still 18 people in the channel
nilay has joined #mlpack
< marcosirc>
So, I can continue..
< marcosirc>
I am Marcos Pividori, a computer science student from Argentina. I have participated in GSoC 13/15, every time was a great experience!
< marcosirc>
This will be my last opportunity as a gsoc student (maximum of 3 times), so I really want to make the most of this!
< keonkim>
wow
< Guest28091>
Wow
< marcosirc>
As most of you, I like playing tennis table too! So, I will join any future mlpack tournament! :)
< marcosirc>
Also, I like playing tennis too, but I am not really good....
< sumedhghaisas>
I like the concept mlpack tournaments :)
< zoq>
me too :)
< rcurtin>
I think I might be eliminated in the first round, I am not very good at table tennis :)
< sumedhghaisas>
we should do that?? like online competitions?? for community bonding.. what say...
< rcurtin>
do they have online table tennis? I am not sure that's quite like the real thing
< rcurtin>
but, if you wanted to organize an activity, I would participate if I was free
< marcosirc>
Haha great!
< nilay>
Hello everyone, My name is Nilay Jain. I am a 3rd year CS undergrad student from BITS Pilani, India. I am doing the " googlenet - we need to go deeper " project. This is the first time i am participating in GSoC. In my free time I like playing football, watching movies and reading.
< keonkim>
Hello, I am Keon Kim (pronounces kee-on Kim) and I will be making “Dataset and experimentation tools” with my mentor Tham.
< rcurtin>
I think BITS Pilani is the most common university for mlpack GSoC students... we've had Mudit in 2013, and Siddharth and Sumedh in 2014
< keonkim>
I am an undergrad student at New York University and am currently working (as an intern) for a startup company that makes software that analyzes financial stuffs using machine learning.
< sumedhghaisas>
I was just going to say that :)
< nilay>
I don't know why but freenode webchat is blocked in my college so I am using a proxy and it gets disconnected sometimes so that is why I have to reconnect so much :(
< keonkim>
This internship will end right before the GSoC starts, so no worries there..
< nilay>
Yes rcurtin we have seminars from our seniors who motivate us to participate in GSoC.
< rcurtin>
keonkim: that sounds very cool, if you happen to work at the same company as a guy named Michael Fox, he is also an mlpack contributor working for a financial startup in NYC
< sumedhghaisas>
we do?? I was not aware of that...
< rcurtin>
although I guess I haven't heard from him in a while, I guess he must be very busy... I understand financial startups take a lot of time :)
< nilay>
Yes sumedhghaisas.. It has been couple of years since we have started doing that. :)
< sumedhghaisas>
I was in Goa campus :) maybe thats the reason...
< tham>
ok, got it, keonkim, let us try our best
< mentekid>
I'm Yannis, student from Thessaloniki, Greece. Hopefully I'll graduate by June, only my thesis remains. I'll do the Locality Sensitive Hashing project, mentored by Ryan.
< nilay>
ok. I am in pilani campus.
< keonkim>
rcurtin: ill keep my eyes open :)
< mentekid>
I was lagging a little behind in school so I've been trying to catch up for the past 1-1.5 years, which means I don't remember what my hobbies are any more :P
< mentekid>
I think I am very typically Greek, I talk too much and too loud, so it's good this isn't skype
< zoq>
:)
< zoq>
Back to this mlpack tournament idea :) Is anybody aware of some mario cart online multiplayer game, I like to beat ryan ^^
< nilay>
zoq: i wanted to ask, you said you could provide ssh access to a good machine. is that possible?
< rcurtin>
haha, I think my mariokart skills are rusty, I will have to practice
< rcurtin>
nilay: if zoq doesn't have a machine, I can probably come up with something, I'll just have to find the right system
< rcurtin>
at GT it was very easy because every system was externally accessible, but at Symantec it's much more difficult to get open ports
< sumedhghaisas>
I also would like to join... though I have never played that game...
< zoq>
nilay: Sure, it isn't as fany as this masterblaster machine, but it should do the job, 64GB RAM and and Corei7
< nilay>
yes because i have a 4gb corei5 here. and i guess this project will require more than that.
< rcurtin>
4GB should be fine, it might be a little slow to compile. I like to do mlpack development on a little chromebook with 1GB RAM and a tiny ARM processor :)
< nilay>
4gb ram*
< rcurtin>
it takes 45 minutes to compile mlpack from scratch...
< sumedhghaisas>
with tests or without??
< rcurtin>
that's just to compile it... I don't know how long it takes to run the tests
< rcurtin>
probably 2 hours...
< tham>
I guess it is without
< zoq>
nilay rcurtin: I think, once we have to train the model, it would come in handy to have a decent machine
< rcurtin>
yeah, definitely if you are doing any hard computation, you are right
< tham>
How many data we want to train?I think we do not need to train the whole data set, it would take a lot of time
< nilay>
and what about training on GPUs can we do that?
< rcurtin>
I think for GPUs the best bet is to use nvblas, the nvidia BLAS library, under Armadillo
< rcurtin>
but I don't know if this will be as fast as something written in CUDA; I haven't played around with it
< rcurtin>
personally I think a lot of CUDA and GPU code is really ugly, and I am not yet sure how to keep mlpack code clean and also support GPUs
< tham>
I will take a look on that, but don't take too many expectation, my machine has some issue with cuda
< zoq>
tham: yeah, I guess, we could test everything on a small subset
< nilay>
oh ok
< rcurtin>
if you think of any way to keep the code clean, we should definitely see if we can figure something nice out
< rcurtin>
I've been writing OpenCL lately, and it's really ugly; I am not sure how it would fit into mlpack nicely
< tham>
Part of the functions call of cuda will crash--after I update it from cuda7.0 to cuda7.5....
< rcurtin>
anyway I have to go get lunch, so I'm headed out for now (I'll be back in an hour or so)
< rcurtin>
thanks everyone for coming to the meeting!
< tham>
I'm tham; I live in Malaysia, apps developer
< tham>
Watch drama, play video game, jogging, study computer vision, coding in my free time
< rcurtin>
I'm here in IRC all the time, I'd encourage everyone to idle in here if possible too, we can discuss mlpack or tournaments or whatever :)
< tham>
I join mlpack community several months ago
< lozhnikov>
Hello! My name is Mikhail Lozhnikov. I am a graduate student at the Moscow State University. I'll work on the implementation of trees for dual-tree algorithms. My interests include computational mathematics, programing. I play classical guitar and piano.
< tham>
mlpack is the first open source project I try to contribute
< tham>
Ryan and Zoq, both of them are friendly and helpful
< rcurtin>
thanks, I am glad I can be helpful :)
< tham>
I learn a lot from this project, thanks a lot
< tham>
I hope I can learn more about ml stuff from GSOC and know more about the community:)
< rcurtin>
classical guitar is really hard, that's very cool
< rcurtin>
anyway, people are waiting on me to come to lunch... talk to you later everybody :)
< Guest28091>
Me too. Ryan and Marcus are both really helpful! And hope learn a lot from this GSoC
< tham>
Have a good lunch
< marcosirc>
Thanks! see you!
< mentekid>
thanks for the intro Ryan! see you later
< keonkim>
:rcurtin thanks for the info, see you later :)
< nilay>
Thank you everyone, great to meet you all :)
< tham>
Great to meet you all too :)
< Guest28091>
Nice to meet you all!
< tham>
I need to go to bed now, see you next time
tham has left #mlpack []
< Guest28091>
I am also leaving for another meeting. See you guys next time~
< keonkim>
nice to meet you all! :)
Guest28091 has quit [Quit: Page closed]
< zoq>
lozhnikov: oh, I always wanted to do vacation in Moscow.
< lozhnikov>
Thanks! See you soon
cache-nez has quit [Quit: Leaving]
< zoq>
lozhnikov: Not sure what's the best month to do that.
< lozhnikov>
Oh, sorry, i write via a phone. And it is slow. I just said goodbye to those, who left this chat:-)
< marcosirc>
I leave too... We stay in contact!
marcosirc has quit [Quit: Page closed]
nilay has quit [Quit: Page closed]
< lozhnikov>
zoq: welcome! But as for me, any old city is better for vacation since there are too much modern buildings. And it causes the lack of beautiful sights.
< zoq>
lozhnikov: So Saint Petersburg is probably the better choice?
< lozhnikov>
St. P. is founded after Moscow. But st.p. is definetely more beautiful!
nilay has joined #mlpack
< zoq>
lozhnikov: hm, do you think it's a good idea to come in November/December? I think it's really cold, but I bet New Year's eve is just awesome in St. P.
< lozhnikov>
It is very cold since there is a sea near St. P.
< zoq>
lozhnikov: I think, in this case it's probably not the best idea :(
Watcher_ has joined #mlpack
< lozhnikov>
zoq: Well.. I can not help you since I have never been there... But all my friends are delighted:-)
< zoq>
lozhnikov: yeah, I talked to some friends that visited some places in Russia and everyone loved it.
< lozhnikov>
I should disconnect for a while. Bye!
< rcurtin>
if I am in MATLAB, the interface should be the same:
< rcurtin>
C = kmeans('input_file', X, 'clusters', 25)
< rcurtin>
or, at least, similar
< rcurtin>
with SWIG, providing bindings like this takes a ton of maintenance
< rcurtin>
personally I'd like to write a better binding generator for mlpack's specific use case
< rcurtin>
but I am not sure exactly what it will look like yet
< rcurtin>
I think it's a bad idea to provide interfaces to every C++ class in mlpack through SWIG, because it would be very easy to use those in a way that makes the code run really slow
< keonkim>
i see
< keonkim>
I am also interested in bindings in Python and R; this will allow more users (tons) to join the community.
< rcurtin>
keonkim: definitely, Python bindings would be useful. but unless we have an automatic binding generator that allows us to keep our own code maintenance overhead down, I think that might be more appropriate as a separate project
< rcurtin>
like RcppMLPACK
< keonkim>
yup I agree
< keonkim>
zoq: thanks for the link!
< zoq>
keonkim: you're welcome!
< keonkim>
rcurtin: and thanks for your insight
ranajn123 has joined #mlpack
sumedhghaisas has quit [Quit: Leaving]
Watcher_ has quit [Ping timeout: 250 seconds]
ranajn123 has quit [Ping timeout: 250 seconds]
Mathnerd314 has quit [Ping timeout: 260 seconds]
Mathnerd314 has joined #mlpack
lozhnik82 has joined #mlpack
lozhnik4 has joined #mlpack
lozhnik82 has quit [Ping timeout: 260 seconds]
lozhnik4 has quit [Ping timeout: 240 seconds]
lozhnik44 has joined #mlpack
lozhnik61 has joined #mlpack
lozhnik44 has quit [Ping timeout: 246 seconds]
Nilabhra has quit [Remote host closed the connection]
lozhnik61 has quit [Quit: -a- IRC for Android 2.1.31]