--- Log opened Sat Jul 16 00:00:25 2011 | ||
-!- blackburn [~blackburn@188.122.238.99] has quit [Quit: Leaving.] | 00:37 | |
@sonney2k | hmmhhh blackburn gone already... | 02:01 |
---|---|---|
-!- f-x` [~user@117.192.203.179] has quit [Ping timeout: 260 seconds] | 04:24 | |
-!- f-x [~user@117.192.203.179] has joined #shogun | 06:24 | |
[1]warpy | morning | 07:20 |
-!- in3xes_ [~in3xes@180.149.49.227] has joined #shogun | 07:22 | |
-!- in3xes [~in3xes@210.212.58.111] has quit [Ping timeout: 240 seconds] | 07:24 | |
-!- f-x [~user@117.192.203.179] has quit [Remote host closed the connection] | 07:48 | |
-!- in3xes [~in3xes@180.149.49.230] has joined #shogun | 08:35 | |
-!- in3xes__ [~in3xes@59.163.196.5] has joined #shogun | 08:40 | |
-!- in3xes1 [~in3xes@180.149.49.227] has joined #shogun | 08:40 | |
-!- in3xes_ [~in3xes@180.149.49.227] has quit [Ping timeout: 240 seconds] | 08:42 | |
-!- in3xes [~in3xes@180.149.49.230] has quit [Ping timeout: 264 seconds] | 08:43 | |
-!- in3xes__ is now known as in3xes | 08:51 | |
-!- gsomix [~gsomix@85.26.235.15] has joined #shogun | 08:58 | |
-!- gsomix [~gsomix@85.26.235.15] has quit [Quit: Ухожу я от вас (xchat 2.4.5 или старше)] | 09:16 | |
-!- alesis_ [~alesis@188.74.87.206] has joined #shogun | 10:00 | |
-!- alesis-novik [~alesis@188.74.87.206] has quit [Ping timeout: 255 seconds] | 10:01 | |
-!- in3xes [~in3xes@59.163.196.5] has quit [Quit: Leaving] | 10:32 | |
CIA-87 | shogun: Shashwat Lal Das master * rca17a15 / src/libshogun/features/StreamingSparseFeatures.h : Added updating of current_num_features for StreamingSparseFeatures. - http://bit.ly/p61GnM | 11:24 |
CIA-87 | shogun: Shashwat Lal Das master * r8f1ed0d / src/libshogun/features/StreamingSparseFeatures.h : Added braces around for loop. - http://bit.ly/o6I0z8 | 11:24 |
CIA-87 | shogun: Shashwat Lal Das master * r563058f / (2 files): Modified bias scale value in SGD. Updated loss function to be the default used in the original svmsgd2. - http://bit.ly/pPkpfd | 11:24 |
CIA-87 | shogun: Soeren Sonnenburg master * r3e6d260 / (3 files in 2 dirs): | 11:24 |
CIA-87 | shogun: Merge pull request #205 from frx/streaming_1 | 11:24 |
CIA-87 | shogun: Added updating of current_num_features for StreamingSparseFeatures. - http://bit.ly/nBNHYa | 11:24 |
-!- blackburn [~blackburn@188.122.238.99] has joined #shogun | 11:34 | |
blackburn | sonney2k: have you seen that msgs from Todd_? | 11:35 |
@sonney2k | blackburn, yes | 11:35 |
blackburn | any candidate? | 11:35 |
blackburn | :) | 11:35 |
@sonney2k | blackburn, we need to recruit new people! | 11:36 |
blackburn | from san francisco? | 11:36 |
blackburn | :D | 11:36 |
blackburn | if you are going to san franciscooo | 11:36 |
blackburn | you should weeeear some flowers in your hair | 11:36 |
@sonney2k | I might attend the GSoC mentors meeting in october but I cannot really travel that much with kids | 11:37 |
@sonney2k | w/o them in this case ... | 11:37 |
blackburn | I see | 11:38 |
CIA-87 | shogun: Soeren Sonnenburg master * r887dcc8 / src/Makefile.template : fix java install - http://bit.ly/nsYZTr | 11:38 |
@sonney2k | blackburn, btw you don't need make install for java (at least currently) | 11:38 |
blackburn | yeah but I have configured it to python_modular and java_modular, so sometimes need to install ;) | 11:39 |
@sonney2k | blackburn, now it works anywyas :) | 11:40 |
blackburn | okay | 11:40 |
blackburn | thanks | 11:40 |
@sonney2k | blackburn, since you are the java expert | 11:41 |
@sonney2k | what are modules in java? | 11:41 |
@sonney2k | I just realized taht modules can implement a certain interface... | 11:42 |
@sonney2k | that is confusing me... | 11:42 |
blackburn | oh okay seems I'm not an expert :D | 11:42 |
blackburn | module - implements? | 11:43 |
@sonney2k | blackburn, well just look at the generated Features.java | 11:44 |
@sonney2k | anyways try to run check.sh and see how far you get... | 11:44 |
@sonney2k | I didn't manage to get the base object to implement the Serializable interface | 11:44 |
@sonney2k | (that statement is missing - rest would be trivial) | 11:44 |
-!- blackburn1 [~blackburn@188.122.238.99] has joined #shogun | 11:45 | |
-!- blackburn [~blackburn@188.122.238.99] has quit [Quit: Leaving.] | 11:45 | |
blackburn1 | sonney2k: I have never been using it :D | 11:46 |
blackburn1 | why you want to? | 11:46 |
[1]warpy | hey guys | 11:48 |
blackburn1 | I thought swig uses packages, no surprisingly modules | 11:48 |
[1]warpy | do you know when splovin is usually around here | 11:48 |
blackburn1 | [1]warpy: hi | 11:48 |
blackburn1 | can't tell for sure | 11:49 |
blackburn1 | well he is from china | 11:49 |
blackburn1 | sonney2k: java modules is a java 7 feature, are you sure you talking about modules? :) | 12:21 |
blackburn1 | what's up with features.java? | 12:24 |
-!- blackburn1 [~blackburn@188.122.238.99] has quit [Read error: No route to host] | 13:01 | |
-!- blackburn [~blackburn@188.122.238.99] has joined #shogun | 13:01 | |
[1]warpy | soeren, can you explain how did you get those svm packages to learn on millions of examples and test on billions ? | 13:28 |
-!- in3xes [~in3xes@180.149.49.230] has joined #shogun | 14:12 | |
-!- Netsplit *.net <-> *.split quits: @mlsec | 14:18 | |
-!- Netsplit over, joins: @mlsec | 14:19 | |
-!- in3xes [~in3xes@180.149.49.230] has quit [Quit: Leaving] | 16:14 | |
-!- in3xes1 [~in3xes@180.149.49.227] has quit [Quit: Leaving] | 16:54 | |
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun | 16:55 | |
-!- mrsrikanth [~mrsrikant@59.92.82.50] has joined #shogun | 17:23 | |
-!- mrsrikanth [~mrsrikant@59.92.82.50] has quit [Quit: Leaving] | 17:56 | |
-!- blackburn [~blackburn@188.122.238.99] has quit [Quit: Leaving.] | 18:25 | |
-!- gsomix [~gsomix@85.26.234.19] has joined #shogun | 19:40 | |
gsomix | hi all | 19:40 |
-!- blackburn [~blackburn@188.122.238.99] has joined #shogun | 20:10 | |
-!- in3xes_ [~in3xes@180.149.49.230] has joined #shogun | 20:34 | |
-!- f-x [~user@117.192.223.132] has joined #shogun | 20:46 | |
-!- gsomix [~gsomix@85.26.234.19] has quit [Ping timeout: 260 seconds] | 21:01 | |
-!- in3xes_ [~in3xes@180.149.49.230] has quit [Ping timeout: 276 seconds] | 22:34 | |
@sonney2k | blackburn, around? | 23:15 |
blackburn | sonney2k: yeap | 23:15 |
@sonney2k | blackburn, hey hey then | 23:15 |
blackburn | :D | 23:15 |
@sonney2k | did you try to run the java examples? | 23:15 |
blackburn | yes, I tried | 23:16 |
blackburn | some fails, some ok | 23:16 |
blackburn | I had a question | 23:16 |
blackburn | minimal test fails, and it's code looks a bit strange | 23:16 |
@sonney2k | classifier_libsvm_minimal_modular ? | 23:17 |
blackburn | minimal_test.java | 23:17 |
@sonney2k | remove this | 23:18 |
blackburn | another question: kernel pca | 23:19 |
blackburn | why it is commented now? | 23:19 |
@sonney2k | no idea | 23:20 |
@sonney2k | shouldn't be | 23:20 |
CIA-87 | shogun: Sergey Lisitsyn master * r4d12457 / examples/undocumented/java_modular/minimal_test.java : Removed unnecessary java example minimal_test - http://bit.ly/nXRaWz | 23:20 |
blackburn | ok it was since we have shogun on github | 23:21 |
blackburn | I am just asking because you wrote it ;) | 23:21 |
@sonney2k | please fix then :) | 23:22 |
blackburn | btw was it working? | 23:22 |
blackburn | I mean have you tested it | 23:22 |
@sonney2k | blackburn, ahh the computation yes - the apply not yet | 23:22 |
blackburn | what is the issue with apply? | 23:23 |
@sonney2k | I didn't implement it yet... | 23:24 |
blackburn | ah okay, can I leave it now, may be some day you will? ;) | 23:25 |
@sonney2k | I would prefer you doing it :) | 23:25 |
@sonney2k | I can send you some matlab code from vojtech for this | 23:25 |
blackburn | hmm yeah, it would be ok | 23:25 |
blackburn | I just have no deep comprehension of (k)pca | 23:26 |
blackburn | strange for me, who are doing dimreduction, yes :D | 23:26 |
bettyboo | <:*) | 23:26 |
blackburn | I'd like to focus on it after HLLE (or in background) | 23:28 |
@sonney2k | sure sure | 23:29 |
blackburn | vojtech is a big fan of matlab ? | 23:30 |
blackburn | looks like he is doing everything in matlab | 23:30 |
@sonney2k | blackburn, he has that big matlab toolbox | 23:30 |
@sonney2k | err wrote | 23:30 |
@sonney2k | not has | 23:30 |
blackburn | which? | 23:30 |
@sonney2k | http://cmp.felk.cvut.cz/cmp/software/stprtool/index.html | 23:31 |
@sonney2k | just look at methods... | 23:31 |
blackburn | oh matlabby shogun | 23:32 |
blackburn | hey it is HUGE | 23:33 |
@sonney2k | blackburn, btw I have an idea for another small task | 23:33 |
@sonney2k | probabilistic outputs for svm | 23:33 |
@sonney2k | that means sigmoid fit | 23:33 |
@sonney2k | or histogram based | 23:33 |
@sonney2k | one could even borrow libsvm's code for the sigmoid fit :) | 23:34 |
bettyboo | !! | 23:34 |
blackburn | I was persuading gsomix to attach C5.0 | 23:35 |
blackburn | sonney2k: have you any ideas how it could be done? | 23:35 |
@sonney2k | C5.0? | 23:35 |
blackburn | yes | 23:36 |
@sonney2k | not really. I havent' looked at their code yet | 23:36 |
blackburn | oh you will be astonished :D | 23:36 |
@sonney2k | what I would do is create a new folder classifier/decisiontree | 23:37 |
@sonney2k | and then move all of c5.0* sources in there and then | 23:37 |
blackburn | it should be some kind of wrapper or so | 23:37 |
@sonney2k | remove the main* stuff | 23:37 |
blackburn | just calling C5.0 things | 23:37 |
@sonney2k | and modify the algorithm to use shogun's feature representation | 23:37 |
@sonney2k | so not a really a wrapper | 23:38 |
@sonney2k | (we did similar things with all the svm's) | 23:38 |
blackburn | cons of this way is that we have to modify that code | 23:38 |
blackburn | not just attach it somehow | 23:38 |
@sonney2k | blackburn, yes - but there is no way around this | 23:39 |
@sonney2k | otherwise we would have to copy the features and have everything twice in memory | 23:39 |
@sonney2k | blackburn, of course one can start like this and do the required modifications later | 23:40 |
blackburn | you should take a look on C5.0 code, it is something I can't describe | 23:40 |
@sonney2k | blackburn, it is from a different century... | 23:41 |
blackburn | something like DSL for C :D | 23:42 |
@sonney2k | blackburn, I mean the author is already retired by now... | 23:44 |
@sonney2k | it is not too bad though | 23:45 |
@sonney2k | (you haven't had a look at svmlight's source code yet :) | 23:45 |
bettyboo | strange :*) | 23:45 |
blackburn | who is author of C5? | 23:45 |
@sonney2k | blackburn, what are the *.i files? | 23:45 |
@sonney2k | Ross Quinlan | 23:46 |
blackburn | no idea, some defines here | 23:46 |
@sonney2k | http://www.rulequest.com/Personal/ | 23:46 |
blackburn | I didn't know he is retired | 23:47 |
@sonney2k | blackburn, what is needed then is to create a wrapper class C50 | 23:47 |
@sonney2k | well he got his PhD in 1968 ... | 23:47 |
blackburn | huh | 23:48 |
f-x | sonney2k: hey, where do you think it would be appropriate to define all the loss functions together? | 23:56 |
@sonney2k | f-x, I think this could be a new directory | 23:57 |
@sonney2k | losses | 23:57 |
@sonney2k | or loss | 23:57 |
f-x | sonney2k: and then something like set_loss_fuction in the classifier api? | 23:59 |
f-x | *function | 23:59 |
--- Log closed Sun Jul 17 00:00:24 2011 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!