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/
na1taneja2821 has quit [Quit: Page closed]
ank_95_ has quit [Quit: Connection closed for inactivity]
agobin has quit [Quit: Connection closed for inactivity]
toshad has quit [Quit: Connection closed for inactivity]
wasiq has quit [Ping timeout: 260 seconds]
decltypeme has joined #mlpack
< decltypeme>
Hi, I wanted to know your feedback about employing multiple classifier systems as a project for GSoC... does it look promising? Are there mentors willing to supervise such a project?
< rcurtin>
decltypeme: what do you mean by multiple classifier systems? do you mean multiple different classifiers (like implement decision trees, SVMs, etc.), or do you mean something else?
< decltypeme>
I mean fusion of decisions generated by classifiers, like we have a KNN, and a naive bayes classifiers for example, then we combine [fuse] the results to obtain a collective behaviour of the classifiers,
< decltypeme>
simplest ways of fusion maybe weighted averages or probability multiplication. However, complex ensemble methods use techniques to obtain variability in children models [at the time of model building]
< rcurtin>
oh so ensembling
< rcurtin>
yeah, that would be cool
< decltypeme>
yes, ensemble methods in classification
< rcurtin>
I've been working on a random forest built of hoeffding trees, but that is not as flexible as what you just proposed
< rcurtin>
I think it would be interesting, especially if it was generic enough to take any classifier
< decltypeme>
yes, exactly, one way of implementation is to deal with the underlying classifiers as blackboxes [that is just one way of it though]
< rcurtin>
yeah
< rcurtin>
I could be interested in mentoring that
< rcurtin>
if you decide to prepare an application, please be sure to detail (among other things) the proposed API, because that's an important part that might need to be discussed a bit
< decltypeme>
Let me write a proposal and get back to you by Monday AoE
< rcurtin>
I can't guarantee that I will have time to look at it on Monday, but I will try to look when I have a chance
< decltypeme>
Sure, whenever you have time
< decltypeme>
So the main [technical] points are how the API of the ensemble model will look like and how it will interface with currently implemented classifiers as well as some literature for guidance, anything else you suggest?
< decltypeme>
ls
< decltypeme>
sorry the last message was to my terminal :D
Nilabhra has joined #mlpack
ank_95_ has joined #mlpack
dd_ has quit [Ping timeout: 252 seconds]
< rcurtin>
decltypeme: that sounds like a good start, don't forget the timeline and to describe the tests you'll write to make sure that it works
< decltypeme>
Do you think the re-factoring work to establish a general class of methods and attributes for the current classifiers in mlpack is too much to be a small module at the beginning of the project?
< rcurtin>
that depends on how much refactoring is necessary to make everything work
< rcurtin>
for the most part the mlpack classifiers kind of implement the same API
< rcurtin>
with a Train() and a Classify() function
chrishenx has quit [Remote host closed the connection]
< decltypeme>
Another question, is MAP adaptation currently implemented for GMMs?
< rcurtin>
I don't think that it is
< decltypeme>
I think it may be another idea, I have broken down my ideas to these two. What do you think?
ranjan123_ has joined #mlpack
Nilabhra has quit [Ping timeout: 252 seconds]
< rcurtin>
decltypeme: sure, either of those would definitely be interesting
< rcurtin>
I'm not familiar with MAP adaptation, so if you submit a proposal for that, please do be sure to describe it well
Nilabhra has joined #mlpack
Awcrr has joined #mlpack
ranjan123_ has quit [Quit: Page closed]
Yvenk has joined #mlpack
Awcrr has quit [Remote host closed the connection]
Awcrr has joined #mlpack
Awcrr_ has joined #mlpack
Awcrr has quit [Ping timeout: 250 seconds]
Awcrr_ has quit [Client Quit]
skon46 has joined #mlpack
mentekid has joined #mlpack
agobin has joined #mlpack
Nilabhra has quit [Remote host closed the connection]
ranjan123 has quit [Quit: Page closed]
ranjan123 has joined #mlpack
Yvenk has quit [Ping timeout: 252 seconds]
wasiq has joined #mlpack
Awcrr has joined #mlpack
wasiq has quit [Ping timeout: 268 seconds]
Awcrr has quit [Remote host closed the connection]
aacr has joined #mlpack
Rishabh_ has quit [Remote host closed the connection]
< saurabh__>
okay thanks!! will go through that and any more suggestion ..... (something to do after that)
Awcrr_ has joined #mlpack
Awcrr__ has quit [Ping timeout: 250 seconds]
Awcrr has quit [Ping timeout: 250 seconds]
< zoq>
saurabh__: Write a good proposal; spend some time thinking about the API of the proposed code and make sure that it fits in with the rest of the codebase.
Awcrr___ has quit [Ping timeout: 250 seconds]
skon46 has joined #mlpack
Awcrr has joined #mlpack
Awcrr_ has quit [Ping timeout: 250 seconds]
Bartek has quit [Remote host closed the connection]
Awcrr_ has joined #mlpack
Awcrr has quit [Ping timeout: 250 seconds]
Nilabhra has joined #mlpack
toshad has joined #mlpack
Awcrr_ has quit []
saurabh__ has quit [Quit: Page closed]
Awcrr has joined #mlpack
Nilabhra has quit [Remote host closed the connection]
Awcrr has quit [Read error: Connection reset by peer]
Awcrr has joined #mlpack
Awcrr_ has joined #mlpack
Awcrr_ has quit [Client Quit]
Awcrr_ has joined #mlpack
Awcrr has quit [Ping timeout: 250 seconds]
Awcrr_ is now known as Awcrr
aacr has quit [Quit: Page closed]
christie has joined #mlpack
< wasiq>
rcurtin, hey I'm working on a proposal for the frank-wolfe algorithm implementation,I've a problem with describing the API that i plan to implement. What exactly are you looking for here?
< rcurtin>
wasiq: I'm looking for the interface that you plan to implement for your optimizer, so I can look over that interface and make sure it will fit with the rest of the mlpack design
< rcurtin>
to be specific for the optimizers, I am looking for an optimizer that has a FunctionType as a template parameter and implements the Optimize() method like the other optimizers
< rcurtin>
past that the details of implementation are a little bit less important
uzipaz has joined #mlpack
travis-ci has joined #mlpack
< travis-ci>
mlpack/mlpack#657 (master - 74d499a : marcus): The build has errored.
johnnytorres83 has quit [Ping timeout: 252 seconds]
vineet has quit [Ping timeout: 268 seconds]
Awcrr has quit [Quit: Leaving...]
wasiq has quit [Ping timeout: 276 seconds]
< ranjan123_>
anyone out there to help me understand this question.
< ranjan123_>
"What areas of machine learning are you familiar with ? "
< ranjan123_>
here "areas of machine learning
< ranjan123_>
"
< ranjan123_>
means what ?
< ranjan123_>
can anyone give me some example ?
rtriangle has joined #mlpack
< rtriangle>
Hello everyone!
< rtriangle>
I am interested in project Dataset and experimentation tools. May I ask if it is still open?
kirizaki has joined #mlpack
< zoq>
ranjan123_: natural language processing, computer vision, a more detailed answer is also possible such as 'familiar with fast nearest neighbor search' ...
< ranjan123_>
hmm. got it @zoq. thank you
< zoq>
rtriangle: No decisions will be made until the application phase ends, so yes the project is still open.
< rtriangle>
thank you)
rtriangle has quit [Quit: Page closed]
christie has quit [Quit: Page closed]
wasiq has joined #mlpack
vineet has joined #mlpack
tsathoggua has joined #mlpack
tbskmmm has quit [Ping timeout: 252 seconds]
sumedhghaisas has joined #mlpack
< wasiq>
rcurtin, got it. Also you mention changes to files such as mlpack_softmax_regression, mlpack_logistic_regression, so that they could use the FW algo.
< wasiq>
rcurtin, would it be feasible for us to add it into summer time frame? IMO it shouldn’t take a lot of time.
< rcurtin>
yeah, changing the executables should not take a long time at all
< rcurtin>
like a day maybe, it's literally just modifying the list of accepted optimizers and writing code to handle the FW optimizer
tbskmmm has joined #mlpack
kirizaki has quit [Quit: Page closed]
< wasiq>
rcurtin, yup understood.Thanks.
mozin has joined #mlpack
tbskmmm has quit [Ping timeout: 240 seconds]
sumedhghaisas has quit [Ping timeout: 240 seconds]
vineet has quit [Ping timeout: 248 seconds]
Bartek has joined #mlpack
mozin has quit [Ping timeout: 252 seconds]
Bartek has quit [Ping timeout: 244 seconds]
trevor has joined #mlpack
trevor is now known as Guest72419
uzipaz has quit [Ping timeout: 252 seconds]
toshad has quit [Quit: Connection closed for inactivity]
Guest72419 has quit [Ping timeout: 248 seconds]
Guest72419 has joined #mlpack
Guest72419 has quit [Ping timeout: 244 seconds]
ranjan123_ has quit [Ping timeout: 252 seconds]
tester_tester has joined #mlpack
tester_tester has quit [Client Quit]
Bartek has joined #mlpack
sumedhghaisas has joined #mlpack
sumedhghaisas has quit [Remote host closed the connection]