--- Log opened Tue Jul 12 00:00:09 2011 | ||
-!- in3xes_ [~in3xes@180.149.49.227] has joined #shogun | 06:42 | |
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 250 seconds] | 06:45 | |
-!- [1]warpy [~warpy@bzq-79-181-19-104.red.bezeqint.net] has joined #shogun | 06:49 | |
-!- todun [~todun@c-68-80-95-42.hsd1.pa.comcast.net] has joined #shogun | 07:18 | |
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun | 08:04 | |
-!- in3xes_ [~in3xes@180.149.49.227] has quit [Ping timeout: 240 seconds] | 08:06 | |
@sonney2k | [1]warpy, still there? | 08:39 |
---|---|---|
@sonney2k | I think you don't need those loadlibrary things | 08:39 |
@sonney2k | IMHO csharp will do that automagically | 08:39 |
@sonney2k | [1]warpy, note however that currently none of the *.cs examples compiles - looks like there is something wrong in daniels typemaps | 08:40 |
-!- in3xes_ [~in3xes@180.149.49.230] has joined #shogun | 08:42 | |
-!- todun [~todun@c-68-80-95-42.hsd1.pa.comcast.net] has quit [Quit: :q] | 09:26 | |
-!- in3xes_ [~in3xes@180.149.49.230] has quit [Remote host closed the connection] | 09:53 | |
-!- [1]warpy [~warpy@bzq-79-181-19-104.red.bezeqint.net] has quit [Ping timeout: 255 seconds] | 10:14 | |
-!- VojtechFranc [~quassel@2001:718:2:1612:221:6aff:fe58:151a] has joined #shogun | 10:14 | |
--- Log closed Tue Jul 12 10:24:22 2011 | ||
--- Log opened Tue Jul 12 10:24:30 2011 | ||
-!- shogun-i1clog [~shogun@7nn.de] has joined #shogun | 10:24 | |
-!- Irssi: #shogun: Total of 10 nicks [2 ops, 0 halfops, 0 voices, 8 normal] | 10:24 | |
-!- Irssi: Join to #shogun was synced in 9 secs | 10:24 | |
-!- Netsplit *.net <-> *.split quits: shogun-irclog | 10:29 | |
-!- blackburn [~blackburn@188.122.238.99] has joined #shogun | 10:52 | |
-!- in3xes_ [~in3xes@180.149.49.227] has joined #shogun | 11:22 | |
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 240 seconds] | 11:25 | |
@sonney2k | alesis-novik, VojtechFranc any status updates? I haven't seen yours this week? | 11:28 |
-!- in3xes1 [~in3xes@180.149.49.227] has joined #shogun | 11:29 | |
blackburn | midterm evaluation took me 5 minutes hehe | 11:30 |
* sonney2k too | 11:32 | |
* sonney2k hopes that blackburn didn't mention vodka or anything alike :-) | 11:33 | |
-!- in3xes_ [~in3xes@180.149.49.227] has quit [Ping timeout: 258 seconds] | 11:33 | |
blackburn | :D | 11:34 |
blackburn | yeah it was hard but I didn't | 11:34 |
-!- in3xes_ [~in3xes@180.149.49.227] has joined #shogun | 11:40 | |
-!- blackburn1 [~blackburn@188.122.238.99] has joined #shogun | 11:41 | |
-!- in3xes1 [~in3xes@180.149.49.227] has quit [Ping timeout: 264 seconds] | 11:44 | |
-!- in3xes1 [~in3xes@180.149.49.227] has joined #shogun | 11:45 | |
-!- blackburn [~blackburn@188.122.238.99] has quit [Ping timeout: 255 seconds] | 11:45 | |
-!- in3xes_ [~in3xes@180.149.49.227] has quit [Ping timeout: 246 seconds] | 11:48 | |
-!- heiko [~heiko@main.uni-duisburg.de] has joined #shogun | 11:50 | |
-!- in3xes_ [~in3xes@180.149.49.227] has joined #shogun | 12:10 | |
-!- in3xes1 [~in3xes@180.149.49.227] has quit [Ping timeout: 252 seconds] | 12:14 | |
@sonney2k | heiko, did you find the bug? | 12:20 |
heiko | sonney2k, hi | 12:20 |
heiko | currently testing from sratch again :( | 12:21 |
heiko | so many possibilities | 12:21 |
heiko | so many changes | 12:21 |
heiko | hope to get it fixed today | 12:21 |
@sonney2k | heiko, I really always use valgrind for testing and pipe this to some file | 12:21 |
@sonney2k | s/this/output/ | 12:21 |
@sonney2k | then search for shogun in there | 12:21 |
@sonney2k | usually there are unitialized memory reads | 12:22 |
@sonney2k | or delete[]/delete / free mismatches or even accidental del/frees | 12:22 |
heiko | yes, am doing this :) | 12:26 |
bettyboo | yeah | 12:26 |
@sonney2k | heiko, that is the best you can do I am afraid... | 12:29 |
@sonney2k | gdb and SG_PRINTS are the other options... | 12:29 |
-!- in3xes_ [~in3xes@180.149.49.227] has quit [Ping timeout: 260 seconds] | 12:35 | |
-!- blackburn1 [~blackburn@188.122.238.99] has quit [Quit: Leaving.] | 12:39 | |
-!- in3xes_ [~in3xes@180.149.49.227] has joined #shogun | 12:47 | |
-!- in3xes1 [~in3xes@180.149.49.227] has joined #shogun | 13:00 | |
-!- in3xes_ [~in3xes@180.149.49.227] has quit [Ping timeout: 240 seconds] | 13:03 | |
-!- in3xes_ [~in3xes@180.149.49.227] has joined #shogun | 14:02 | |
-!- in3xes1 [~in3xes@180.149.49.227] has quit [Ping timeout: 240 seconds] | 14:05 | |
heiko | ah, just found a bunch of bugs :) | 14:07 |
-!- in3xes1 [~in3xes@180.149.49.227] has joined #shogun | 14:08 | |
-!- in3xes_ [~in3xes@180.149.49.227] has quit [Ping timeout: 240 seconds] | 14:11 | |
@sonney2k | heiko, and? does it help? | 14:52 |
heiko | sonney2k, yes some lower parts are working again | 14:53 |
heiko | making my way up to the top | 14:53 |
heiko | I implemented a special feature in DynamicObjectArray that deleted elements unwanted :) | 14:53 |
@sonney2k | 'unwanted'? | 14:54 |
@sonney2k | as in delete_unwanted_elements :D | 14:54 |
heiko | yeah ;D | 14:54 |
heiko | well elements that you just put in but that the array did not want | 14:55 |
@sonney2k | ? | 15:01 |
@sonney2k | heiko, still isn't really clear to me... | 15:02 |
-!- [1]warpy [~warpy@bzq-79-181-19-104.red.bezeqint.net] has joined #shogun | 15:13 | |
heiko | sonney2k, the array did eat up elements, this was not intended. | 15:34 |
heiko | a bug | 15:34 |
[1]warpy | hi sonney2k saw your msg from earlier | 15:34 |
bettyboo | ho | 15:34 |
@sonney2k | heiko, you mean the array didn't shrink? | 15:37 |
heiko | no, just a bug | 15:38 |
@sonney2k | I see | 15:38 |
@sonney2k | ok | 15:38 |
heiko | wrong element deleted | 15:38 |
-!- VojtechFranc [~quassel@2001:718:2:1612:221:6aff:fe58:151a] has quit [Quit: No Ping reply in 180 seconds.] | 15:48 | |
-!- VojtechFranc [~quassel@cmpgw-27.felk.cvut.cz] has joined #shogun | 15:50 | |
-!- heiko [~heiko@main.uni-duisburg.de] has quit [Ping timeout: 258 seconds] | 16:05 | |
[1]warpy | sonney2k, is there a way to issue a pull via the git website ? | 16:06 |
@sonney2k | http://www.google.com/search?q=github+pull+request | 16:06 |
[1]warpy | so basically i push it first | 16:09 |
[1]warpy | and then i pull (ask you to look at it) | 16:09 |
@sonney2k | yes | 16:12 |
[1]warpy | okay, so can i push stuff or do i need permission | 16:19 |
-!- blackburn [~blackburn@188.122.238.99] has joined #shogun | 16:20 | |
-!- in3xes1 [~in3xes@180.149.49.227] has quit [Quit: Leaving] | 16:24 | |
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun | 16:25 | |
@sonney2k | you can push as much as you want :) | 16:31 |
blackburn | i'd like to push some cats into shogun | 16:36 |
[1]warpy | sonney2k, what is the problem with the type map can you possibly email me the errors ? | 16:41 |
[1]warpy | and also to daniel ? | 16:41 |
@sonney2k | blackburn, shogun takes out is sword and cuts the cat into nice sashimi pieces to be served for dinner | 16:43 |
blackburn | sonney2k: have you vietnamese ancestors? :D strange tastes ;) | 16:45 |
bettyboo | ;D | 16:45 |
-!- blackburn1 [~blackburn@188.122.238.99] has joined #shogun | 16:50 | |
-!- blackburn [~blackburn@188.122.238.99] has quit [Ping timeout: 255 seconds] | 16:51 | |
[1]warpy | sonney2k, i forked the project, i have no idea how to push/commit | 16:53 |
[1]warpy | from the web obviously | 16:55 |
[1]warpy | not command line | 16:55 |
blackburn1 | just make changes, pull origin, commit it, push it to your fork and press pull request button on web | 16:56 |
-!- f-x [~user@117.192.220.22] has joined #shogun | 16:57 | |
-!- f-x [~user@117.192.220.22] has quit [Remote host closed the connection] | 17:10 | |
-!- VojtechFranc [~quassel@cmpgw-27.felk.cvut.cz] has quit [Remote host closed the connection] | 17:26 | |
-!- heiko [~heiko@main.uni-duisburg.de] has joined #shogun | 18:08 | |
heiko | sonney2k, are you there? | 18:11 |
-!- heiko [~heiko@main.uni-duisburg.de] has quit [Quit: Leaving.] | 18:26 | |
-!- blackburn1 [~blackburn@188.122.238.99] has quit [Quit: Leaving.] | 19:02 | |
-!- blackburn [~blackburn@188.122.238.99] has joined #shogun | 19:02 | |
-!- [1]warpy [~warpy@bzq-79-181-19-104.red.bezeqint.net] has quit [Quit: HydraIRC -> http://www.hydrairc.com <- Po-ta-to, boil em, mash em, stick em in a stew.] | 19:12 | |
-!- in3xes is now known as gsoc | 19:13 | |
-!- gsoc is now known as group | 19:14 | |
-!- group is now known as gsoc | 19:14 | |
-!- gsoc is now known as in3xes | 19:15 | |
-!- f-x [~user@117.192.220.22] has joined #shogun | 19:38 | |
-!- in3xes_ [~in3xes@180.149.49.230] has joined #shogun | 20:10 | |
-!- blackburn [~blackburn@188.122.238.99] has quit [Quit: Leaving.] | 20:19 | |
-!- blackburn [~blackburn@188.122.238.99] has joined #shogun | 20:32 | |
@sonney2k | f-x, so how is it going? | 21:02 |
@sonney2k | did you start working on liblinear/sgd-qn? | 21:03 |
@sonney2k | f-x, if not I have one more proposal :) | 21:03 |
bettyboo | great | 21:03 |
@sonney2k | f-x, you could download the original sgd / sgd-qn and run it on this subset of data that you have | 21:03 |
f-x | sonney2k: started, but i'd love to hear your proposal :) | 21:03 |
@sonney2k | these can read svmlight file format | 21:04 |
@sonney2k | (the format the python script of the large scale learnign challenge produces) | 21:04 |
@sonney2k | so you can directly run them and check if they get very similar results | 21:04 |
f-x | ok | 21:05 |
@sonney2k | f-x, but it is great that you started already | 21:05 |
f-x | i ran liblinear on a small subset of data, but it always ended up reaching max iterations | 21:05 |
@sonney2k | f-x, I am very excited about this... | 21:05 |
f-x | (the original liblinear from shogun) | 21:05 |
@sonney2k | f-x, on alpha even? | 21:05 |
@sonney2k | f-x, which solver did you choose? | 21:06 |
f-x | sonney2k: everything default | 21:06 |
@sonney2k | (for liblinear) | 21:06 |
@sonney2k | f-x, then of course :) | 21:06 |
f-x | which is the fastest one? | 21:06 |
@sonney2k | f-x, you need to use L2R L1LOSS | 21:06 |
@sonney2k | read L_2 norm regularized SVM-w with L_1-norm loss term | 21:06 |
f-x | ah.. i'll try that.. meanwhile i do have some questions about the online version.. i'll ask you a bit later | 21:07 |
@sonney2k | that is the SVM | 21:07 |
@sonney2k | default in liblinear is IIRC some logistic regression | 21:07 |
f-x | oh.. btw liblinear has come out with a new version which reads from datasets that don't fit in the memory | 21:08 |
f-x | sonney2k: they split the problem into multiple blocks, and process one block at a time | 21:08 |
@sonney2k | f-x, I know | 21:08 |
@sonney2k | f-x, but it is not in liblinear directly AFAIK it is some addon | 21:09 |
f-x | they call it the 'block' version iirc | 21:09 |
@sonney2k | f-x, yes but it is some extra c++ code that just uses liblinear or? | 21:09 |
f-x | sonney2k: they modify the train and predict functions of the original liblinear.. | 21:10 |
f-x | so i'd consider it to be a standalone package | 21:10 |
@sonney2k | I just checked http://www.csie.ntu.edu.tw/~cjlin/liblinear/ | 21:10 |
@sonney2k | so yes indeed | 21:10 |
@sonney2k | it is not in core liblinear | 21:10 |
@sonney2k | f-x, if you have time later - this is something you could do | 21:11 |
f-x | sonney2k: and one more thing - i'm finding it easier to be using DynArray instead of int* etc for liblinear | 21:11 |
f-x | is that okay? | 21:11 |
@sonney2k | int* | 21:11 |
@sonney2k | for what? | 21:11 |
@sonney2k | blackburn, my replacement harddrive just arrived today | 21:11 |
@sonney2k | blackburn, isn't that great news :D | 21:11 |
blackburn | sonney2k: of course it is ;) | 21:12 |
@sonney2k | blackburn, the only sad thing is that the drive's label says Certified Repaired HDD | 21:12 |
@sonney2k | that scares me quite a bit | 21:13 |
@sonney2k | f-x, int* for what? | 21:13 |
blackburn | bastards, I thought it is possible only in russia :D | 21:13 |
f-x | sonney2k: umm.. the y values | 21:13 |
f-x | doesn't liblinear depend on multiple iterations? | 21:13 |
f-x | is convergence guaranteed without this? | 21:14 |
@sonney2k | f-x, not more than SGD | 21:14 |
@sonney2k | f-x, no convergence is not guaranteed | 21:14 |
f-x | because i see it retains some examples and discards others, and runs another iteration on the retained examples | 21:15 |
f-x | i thought that was essential | 21:15 |
blackburn | sonney2k: I started glancing PCACut now. is it PCA *Cut*?? | 21:16 |
@sonney2k | f-x, it is if you want to converge to some epsilon precision | 21:16 |
@sonney2k | f-x, the discarding stuff is 'only' for speeding up | 21:17 |
@sonney2k | not essential in the online variant | 21:17 |
f-x | aah.. that solves most of my worries.. i thought somehow convergence could be guaranteed even through this online method | 21:17 |
@sonney2k | f-x, well with infinite data SGD and liblinear will converge to the same solution :) | 21:18 |
f-x | great.. so the basic mechanism is nearly identical | 21:18 |
f-x | sonney2k: but i'll have to modify some fundamental structures also -- like 'problem' | 21:20 |
f-x | again CDotFeatures -> CStreamingDotFeatures | 21:20 |
f-x | i don't see a way without creating something like online_problem | 21:21 |
@sonney2k | f-x, yeah or do without problem | 21:23 |
f-x | yes - another possibility | 21:23 |
f-x | i guess i'll do it without the problem stuff, that's good | 21:24 |
@sonney2k | f-x, just to be clear: you only need to convert solve_l2r_l1l2_svc | 21:26 |
f-x | sonney2k: right. i've only looked at that for now | 21:26 |
@sonney2k | that is liblinear's svm like training algorithm | 21:26 |
@sonney2k | no more than that | 21:26 |
@sonney2k | ok | 21:26 |
@sonney2k | the rest is not necessary / not so useful | 21:27 |
f-x | ok. so the generic online features loading stuff, then train using solve_l2r_l1l2_svc, and again the generic predict stuff | 21:27 |
f-x | only solve_l2r_l1l2_svc is specific to liblinear | 21:28 |
@sonney2k | f-x, yeah so just copy that function to OnlineLibLinear's train() function and modify it such that it can use streamingfeatures | 21:35 |
@sonney2k | that's it | 21:35 |
@sonney2k | StreamingDotFeatures that is | 21:35 |
f-x | yeah.. that way is much more compact.. thanks | 21:36 |
@sonney2k | blackburn, I named it this way because it is doing PCA and could cut off features with low EV's | 21:36 |
@sonney2k | blackburn, but you are the preproc master now | 21:36 |
blackburn | sonney2k: I guess it is time to rename it to PCA - it transforms data, not cut off it | 21:37 |
@sonney2k | f-x, yeah it is not a lot of work when you look at it directly from that angle | 21:37 |
blackburn | or I got wrong? | 21:37 |
@sonney2k | blackburn, as you wish :) | 21:39 |
blackburn | sonney2k: well it transforms data, right? :) | 21:40 |
@sonney2k | blackburn, as I said - you are the master of preprocs now - so just do it - I am fine with either name :) | 21:41 |
blackburn | okay | 21:41 |
* sonney2k bites the dust and works on c# typemaps | 22:08 | |
-!- blackburn [~blackburn@188.122.238.99] has quit [Quit: Leaving.] | 22:16 | |
-!- in3xes_ [~in3xes@180.149.49.230] has quit [Ping timeout: 260 seconds] | 22:40 | |
--- Log closed Wed Jul 13 00:00:14 2011 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!