--- Log opened Mon Sep 23 00:00:11 2013 | ||
-!- gsomix [~gsomix@83.234.54.128] has joined #shogun | 00:14 | |
gsomix | sonimperator, van_solo hey, star wars party? | 00:15 |
---|---|---|
van_solo | gsomix: yeah and we are looking for a chewbacca | 00:15 |
-!- gsomix is now known as gLando | 00:28 | |
gLando | van_solo, not today | 00:28 |
-!- lisitsyn1 [~lisitsyn@fb2-lo1.global63.net] has joined #shogun | 00:32 | |
van_solo | haha nice choice gLando | 00:32 |
van_solo | I don't think we are going to find one | 00:32 |
van_solo | we should just rename the buildbot :p | 00:32 |
gLando | van_solo, yep. buildbot only does that growls and roars | 00:34 |
-!- lisitsyn [~lisitsyn@fb2-lo1.global63.net] has quit [Ping timeout: 256 seconds] | 00:34 | |
-!- lisitsyn1 is now known as lisitsyn | 00:34 | |
-!- HeikoS is now known as Heikowankenobi | 00:48 | |
lisitsyn | Heikowankenobi: oh man you too | 00:48 |
@Heikowankenobi | haha :) | 00:48 |
lisitsyn | HeikoWanHeikobi Heikowankenobi | 00:50 |
van_solo | gLando: will you export LineReader to the modular interfaces? | 01:01 |
gLando | van_solo, not sure. | 01:01 |
van_solo | gLando: It had freed my hands with reading \0 delimited files :) | 01:02 |
gLando | van_solo, ok. | 01:02 |
gLando | van_solo, how urgent is this needed? | 01:03 |
van_solo | gLando: I've found some python specific stuff I'll try those for now | 01:03 |
van_solo | gLando: not really urgent. it was more of a question | 01:03 |
van_solo | gLando: I can do it locally if it comes to that, so no worries :) | 01:04 |
gLando | van_solo, ok. | 01:04 |
gLando | van_solo, you know, I'm guilty of... sealing you in carbonite. so I'll do it for you. | 01:08 |
van_solo | gLando: haha do that and maybe I'll forgive you | 01:09 |
gLando | /usr/lib/gcc/x86_64-linux-gnu/4.7/../../../x86_64-linux-gnu/crt1.o: In function `_start': | 01:51 |
gLando | (.text+0x20): undefined reference to `main' | 01:51 |
gLando | collect2: error: ld returned 1 exit status | 01:51 |
gLando | very nice error | 01:51 |
gLando | while linking converter_jade_bss | 01:52 |
-!- lambday [67157f4c@gateway/web/freenode/ip.103.21.127.76] has joined #shogun | 02:18 | |
lambday | Heikowankenobi: hi | 02:18 |
lambday | :D | 02:18 |
@Heikowankenobi | lambday: hey! :) | 02:18 |
lambday | hahaha | 02:18 |
lambday | nice name :D | 02:18 |
@Heikowankenobi | yeah people are freaking out here! :) | 02:19 |
lambday | Heikowankenobi: I started it as luke :D | 02:19 |
@Heikowankenobi | lambday: haha really? :D | 02:19 |
lambday | yep :D | 02:19 |
gLando | sonimperator, https://github.com/shogun-toolbox/shogun/pull/1651 updated | 02:19 |
lambday | Heikowankenobi: is it okay if I add the histogram later? | 02:20 |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 02:20 | |
shogun-notifier- | shogun: Heiko Strathmann :gp_notebook * 7d24649 / tests/integration/python_modular/tester.py: https://github.com/shogun-toolbox/shogun/commit/7d24649783b153e3648fef55bcd6f592efbff429 | 02:20 |
shogun-notifier- | shogun: suggestion for output in case of failing integration test | 02:20 |
shogun-notifier- | shogun: Heiko Strathmann :gp_notebook * ded9eb2 / doc/ipython-notebooks/gaussian_process/gaussian_process_classification.ipynb: https://github.com/shogun-toolbox/shogun/commit/ded9eb2ebe3d0558d27db09e45396e5614e07b9a | 02:20 |
shogun-notifier- | shogun: initial draft for gp notebook | 02:20 |
@Heikowankenobi | lambday: got other plans for tomorrow? :) | 02:21 |
lambday | Heikowankenobi: yeah... really gotta get into insti stuffs now | 02:21 |
lambday | Heikowankenobi: I'll remove the outputs | 02:21 |
@Heikowankenobi | lambday: I would like to see the thing finished before tomorrow night, but dont worry if you are in trouble | 02:22 |
@Heikowankenobi | lambday: the histogram should be easy to fix though | 02:22 |
@Heikowankenobi | just hist of the samples you already have in there | 02:22 |
@Heikowankenobi | two lines | 02:22 |
@Heikowankenobi | lambday: but yeah not really ultra important | 02:22 |
lambday | Heikowankenobi: yeah once I figure out how to plot it properly its not much work | 02:22 |
lambday | histograms are always confusing to me :-/ | 02:23 |
shogun-notifier- | shogun: Heiko Strathmann :develop * 7d24649 / tests/integration/python_modular/tester.py: https://github.com/shogun-toolbox/shogun/commit/7d24649783b153e3648fef55bcd6f592efbff429 | 02:23 |
shogun-notifier- | shogun: suggestion for output in case of failing integration test | 02:23 |
shogun-notifier- | shogun: Heiko Strathmann :develop * 85eeb68 / tests/integration/python_modular/tester.py: https://github.com/shogun-toolbox/shogun/commit/85eeb689b0fac2c33da68faec10fdb1d35bcbc29 | 02:23 |
shogun-notifier- | shogun: Merge pull request #1658 from karlnapf/develop | 02:23 |
shogun-notifier- | shogun: | 02:23 |
shogun-notifier- | shogun: suggestion for output in case of failing integration test | 02:23 |
lambday | Heikowankenobi: but checking ImportErrors... confused.. | 02:23 |
lambday | Heikowankenobi: should I just guard the whole thing in try catch? | 02:23 |
-!- van_solo [~van51@athedsl-409794.home.otenet.gr] has quit [Quit: Leaving.] | 02:25 | |
@Heikowankenobi | lambday: didnt I give you code? | 02:27 |
@Heikowankenobi | lambday: ah yes, that is what I had in mind, but that is in fact stupid | 02:27 |
@Heikowankenobi | lambday: so lets just install everything on the notebook machine | 02:27 |
@Heikowankenobi | lambday: dont worry about this then, sorry for the confusion | 02:27 |
@Heikowankenobi | histogram plotting is very easy | 02:28 |
@Heikowankenobi | try this: | 02:28 |
lambday | Heikowankenobi: yeah it should be, otherwise just blank outputs in the notebooks won't be useful :( | 02:28 |
@Heikowankenobi | hist(randn(1000)*0.2) | 02:28 |
@Heikowankenobi | and then hist(randn(1000)) | 02:28 |
@Heikowankenobi | automagically puts them in one plot with different colours | 02:28 |
lambday | Heikowankenobi: alright trying | 02:28 |
@Heikowankenobi | lambday: exactly, blanking is not useful | 02:28 |
@Heikowankenobi | lambday: and then legend(["a", "b"]) | 02:31 |
@Heikowankenobi | and then maybe | 02:31 |
@Heikowankenobi | plot([true_logdet, true_logdet], [0,some-value], linewidth=3) for the true value line | 02:31 |
shogun-buildbot | build #1795 of bsd1 - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/1795 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com> | 02:31 |
lambday | Heikowankenobi: wow this works | 02:32 |
lambday | Heikowankenobi: and plotting straight line? | 02:32 |
@Heikowankenobi | lambday: see above | 02:32 |
lambday | oh | 02:32 |
lambday | sorry I missed that | 02:32 |
gLando | nite | 02:33 |
@Heikowankenobi | no worries | 02:33 |
lambday | checking | 02:33 |
@Heikowankenobi | gLando: nite! :) | 02:33 |
lambday | gLando: fernando? | 02:34 |
gLando | lambday, nice try, but no. :) | 02:35 |
lambday | gLando: wiking | 02:35 |
lambday | :D | 02:35 |
lambday | gLando: good night | 02:36 |
shogun-notifier- | shogun: Roman Votyakov :develop * d255abb / / (76 files): https://github.com/shogun-toolbox/shogun/commit/d255abb1d4c562a2ea8bc280b3994a0569ed8ec6 | 02:44 |
shogun-notifier- | shogun: refactor gradient model selection framework | 02:44 |
shogun-notifier- | shogun: Heiko Strathmann :develop * 6ddd47d / / (76 files): https://github.com/shogun-toolbox/shogun/commit/6ddd47d286ebf3ad045dd921eeda2e4ef14ee62d | 02:44 |
shogun-notifier- | shogun: Merge pull request #1640 from votjakovr/feature/gp_refactoring | 02:44 |
shogun-notifier- | shogun: | 02:44 |
shogun-notifier- | shogun: Refactor gradient model selection | 02:44 |
gLando | Heikowankenobi, btw I sent small patch. https://github.com/shogun-toolbox/shogun/pull/1662 And it's green for clang now. | 02:44 |
@Heikowankenobi | gLando: checking | 02:44 |
shogun-notifier- | shogun: Evgeniy Andreev :develop * 27deaee / examples/undocumented/libshogun/converter_jade_bss.cpp: https://github.com/shogun-toolbox/shogun/commit/27deaeeb8291daead04abd4b3df306bc4feca74e | 02:46 |
shogun-notifier- | shogun: fix converter_jade_bss example | 02:46 |
shogun-notifier- | shogun: Heiko Strathmann :develop * 5d1a561 / examples/undocumented/libshogun/converter_jade_bss.cpp: https://github.com/shogun-toolbox/shogun/commit/5d1a5612aaab2c74835ea2373be1aad46302d997 | 02:46 |
shogun-notifier- | shogun: Merge pull request #1662 from gsomix/feature/jade_bss | 02:46 |
shogun-notifier- | shogun: | 02:46 |
shogun-notifier- | shogun: Fix converter_jade_bss example | 02:46 |
gLando | Heikowankenobi, cool! for now I can sleep with no worries... | 02:46 |
@Heikowankenobi | gLando: hehe :) Shogun worries you in your dreams? :D | 02:47 |
@Heikowankenobi | same here, haha ;) | 02:47 |
gLando | Heikowankenobi, sometimes. no more often than girls | 02:49 |
gLando | ok | 02:49 |
@Heikowankenobi | gLando: phew! thats good | 02:49 |
@Heikowankenobi | gLando: speaking of, I am joining mine now :) see you later | 02:49 |
shogun-buildbot | build #1526 of cyg1 - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/cyg1%20-%20libshogun/builds/1526 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com> | 02:55 |
lambday | Heikowankenobi: updated the notebook | 02:57 |
@Heikowankenobi | lambday: wow that was fast! :) | 02:57 |
@Heikowankenobi | lambday: mmmh parsing the diff is like impossible, I will just trust you! :) | 02:59 |
shogun-notifier- | shogun: lambday :develop * 2e242a3 / doc/ipython-notebooks/logdet/logdet.ipynb: https://github.com/shogun-toolbox/shogun/commit/2e242a354911479cf11df2ce76124662dcdbd160 | 02:59 |
shogun-notifier- | shogun: ipython notebook added (log-det) | 02:59 |
shogun-notifier- | shogun: Heiko Strathmann :develop * 301733a / doc/ipython-notebooks/logdet/logdet.ipynb: https://github.com/shogun-toolbox/shogun/commit/301733ade16aa11a9f0a3a9c85c01f875f9afaec | 02:59 |
shogun-notifier- | shogun: Merge pull request #1659 from lambday/feature/log_determinant | 02:59 |
shogun-notifier- | shogun: | 02:59 |
shogun-notifier- | shogun: ipython notebook added (log-det) | 02:59 |
@Heikowankenobi | lambday: could you quickly update the old gist you sent around? I sent that to some people | 03:00 |
-!- lambday [67157f4c@gateway/web/freenode/ip.103.21.127.76] has quit [Ping timeout: 250 seconds] | 03:01 | |
shogun-buildbot | build #1876 of deb3 - modular_interfaces is complete: Failure [failed test python modular test ruby modular] Build details are at http://buildbot.shogun-toolbox.org/builders/deb3%20-%20modular_interfaces/builds/1876 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com> | 03:20 |
shogun-buildbot | build #1527 of cyg1 - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/cyg1%20-%20libshogun/builds/1527 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com> | 03:22 |
shogun-notifier- | shogun: Heiko Strathmann :gp_notebook * d8fe94e / doc/ipython-notebooks/gaussian_process/gaussian_process_classification.ipynb: https://github.com/shogun-toolbox/shogun/commit/d8fe94edd497ae069634d47798eed7c310e08883 | 03:22 |
shogun-notifier- | shogun: initial draft for gp notebook | 03:22 |
shogun-buildbot | build #1877 of deb3 - modular_interfaces is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/deb3%20-%20modular_interfaces/builds/1877 | 03:29 |
-!- travis-ci [~travis-ci@ec2-54-211-149-66.compute-1.amazonaws.com] has joined #shogun | 03:36 | |
travis-ci | [travis-ci] it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/11670381 | 03:36 |
-!- travis-ci [~travis-ci@ec2-54-211-149-66.compute-1.amazonaws.com] has left #shogun [] | 03:36 | |
-!- Heikowankenobi [~heiko@218.4.166.178.rev.vodafone.pt] has quit [Quit: Leaving.] | 03:37 | |
shogun-buildbot | build #1796 of bsd1 - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/1796 blamelist: Evgeniy Andreev <gsomix@gmail.com>, Heiko Strathmann <heiko.strathmann@gmail.com>, lambday <heavensdevil6909@gmail.com>, Roman Votyakov <votjakovr@gmail.com> | 03:45 |
shogun-buildbot | build #1248 of rpm1 - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/rpm1%20-%20libshogun/builds/1248 blamelist: Evgeniy Andreev <gsomix@gmail.com>, Heiko Strathmann <heiko.strathmann@gmail.com>, lambday <heavensdevil6909@gmail.com>, Roman Votyakov <votjakovr@gmail.com> | 03:48 |
-!- sonne|osx_ [~sonne@g225081034.adsl.alicedsl.de] has joined #shogun | 03:57 | |
shogun-buildbot | build #134 of precise - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/precise%20-%20libshogun/builds/134 blamelist: Evgeniy Andreev <gsomix@gmail.com>, Heiko Strathmann <heiko.strathmann@gmail.com>, lambday <heavensdevil6909@gmail.com>, Roman Votyakov <votjakovr@gmail.com> | 03:57 |
shogun-buildbot | build #169 of FC19 - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/FC19%20-%20libshogun/builds/169 blamelist: Evgeniy Andreev <gsomix@gmail.com>, Heiko Strathmann <heiko.strathmann@gmail.com>, lambday <heavensdevil6909@gmail.com>, Roman Votyakov <votjakovr@gmail.com> | 03:58 |
shogun-buildbot | build #168 of FCRH - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/FCRH%20-%20libshogun/builds/168 blamelist: Evgeniy Andreev <gsomix@gmail.com>, Heiko Strathmann <heiko.strathmann@gmail.com>, lambday <heavensdevil6909@gmail.com>, Roman Votyakov <votjakovr@gmail.com> | 03:58 |
-!- sonne|osx [~sonne@g225081145.adsl.alicedsl.de] has quit [Ping timeout: 248 seconds] | 03:59 | |
-!- sonne|osx_ is now known as sonne|osx | 03:59 | |
shogun-buildbot | build #464 of nightly_all is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/nightly_all/builds/464 | 04:05 |
shogun-buildbot | build #1528 of cyg1 - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/cyg1%20-%20libshogun/builds/1528 blamelist: Evgeniy Andreev <gsomix@gmail.com>, Heiko Strathmann <heiko.strathmann@gmail.com>, lambday <heavensdevil6909@gmail.com>, Roman Votyakov <votjakovr@gmail.com> | 04:15 |
shogun-buildbot | build #33 of clang34 - static analysis is complete: Failure [failed analyse] Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20static%20analysis/builds/33 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com> | 04:21 |
shogun-buildbot | build #19 of clang34 - undefined behaviour analysis is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20undefined%20behaviour%20analysis/builds/19 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com> | 04:31 |
shogun-buildbot | build #23 of clang34 - thread analysis is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20thread%20analysis/builds/23 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com> | 04:48 |
shogun-buildbot | build #27 of debian wheezy - memcheck is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/debian%20wheezy%20-%20memcheck/builds/27 | 04:52 |
-!- travis-ci [~travis-ci@ec2-54-211-149-66.compute-1.amazonaws.com] has joined #shogun | 04:54 | |
travis-ci | [travis-ci] it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/11670413 | 04:54 |
-!- travis-ci [~travis-ci@ec2-54-211-149-66.compute-1.amazonaws.com] has left #shogun [] | 04:54 | |
-!- travis-ci [~travis-ci@ec2-54-211-149-66.compute-1.amazonaws.com] has joined #shogun | 05:41 | |
travis-ci | [travis-ci] it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/11670733 | 05:41 |
-!- travis-ci [~travis-ci@ec2-54-211-149-66.compute-1.amazonaws.com] has left #shogun [] | 05:41 | |
-!- Cheng [80fa05f8@gateway/web/freenode/ip.128.250.5.248] has joined #shogun | 05:52 | |
Cheng | I've got a newbie question: I can't seem to use cmake properly, using the instructions in README.cmake | 05:53 |
Cheng | I get the error: | 05:53 |
Cheng | CMake Error: Attempt to add a custom rule to output "/Users/cheng.ong/External/shogun/src/interfaces/python_modular/abstract_types_extension.i.rule" which already has a custom rule. | 05:53 |
Cheng | And so on... | 05:54 |
Cheng | I'm on the develop branch of git. | 05:54 |
Cheng | cmake version 2.8.11.2 on OSX | 06:00 |
wiking | Cheng: remove the symlinks in /Users/cheng.ong/External/shogun/src/interfaces/python_modular/ | 06:21 |
wiking | but only thr symlinks | 06:22 |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 06:22 | |
-!- travis-ci [~travis-ci@ec2-54-243-23-104.compute-1.amazonaws.com] has joined #shogun | 06:27 | |
travis-ci | [travis-ci] it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/11670770 | 06:27 |
-!- travis-ci [~travis-ci@ec2-54-243-23-104.compute-1.amazonaws.com] has left #shogun [] | 06:27 | |
-!- zxtx_ [~zv@209-6-202-142.c3-0.smr-ubr2.sbo-smr.ma.cable.rcn.com] has joined #shogun | 06:38 | |
Cheng | wiking: Thanks for the suggestion, but I don't see any symlinks | 06:52 |
shogun-buildbot | build #34 of clang34 - static analysis is complete: Failure [failed analyse] Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20static%20analysis/builds/34 blamelist: Evgeniy Andreev <gsomix@gmail.com>, Heiko Strathmann <heiko.strathmann@gmail.com>, Roman Votyakov <votjakovr@gmail.com> | 06:56 |
gLando | good morning | 07:05 |
-!- votjakovr [~votjakovr@host-46-241-3-209.bbcustomer.zsttk.net] has joined #shogun | 07:29 | |
-!- sonne|osx [~sonne@g225081034.adsl.alicedsl.de] has quit [Quit: sonne|osx] | 07:29 | |
shogun-buildbot | build #35 of clang34 - static analysis is complete: Failure [failed analyse] Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20static%20analysis/builds/35 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, lambday <heavensdevil6909@gmail.com> | 07:37 |
wiking | Cheng: /Users/cheng.ong/External/shogun/src/interfaces/python_modular/abstract_types_extension.i.rule is a symlink to /Users/cheng.ong/External/shogun/src/interfaces/modular/abstract_types_extension.i.rule | 07:48 |
shogun-buildbot | build #20 of clang34 - undefined behaviour analysis is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20undefined%20behaviour%20analysis/builds/20 blamelist: Evgeniy Andreev <gsomix@gmail.com>, Heiko Strathmann <heiko.strathmann@gmail.com>, lambday <heavensdevil6909@gmail.com>, Roman Votyakov | 07:49 |
shogun-buildbot | <votjakovr@gmail.com> | 07:49 |
-!- travis-ci [~travis-ci@ec2-54-211-149-66.compute-1.amazonaws.com] has joined #shogun | 07:51 | |
travis-ci | [travis-ci] it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/11671070 | 07:51 |
-!- travis-ci [~travis-ci@ec2-54-211-149-66.compute-1.amazonaws.com] has left #shogun [] | 07:51 | |
Cheng | 400L-97169:python_modular cheng.ong$ pwd /Users/cheng.ong/External/shogun/src/interfaces/python_modular 400L-97169:python_modular cheng.ong$ ls *.rule ls: *.rule: No such file or directory | 07:55 |
Cheng | Sorry, missing newlines. Basically I don't have any files with .rule extensions | 07:56 |
shogun-buildbot | build #24 of clang34 - thread analysis is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20thread%20analysis/builds/24 blamelist: Evgeniy Andreev <gsomix@gmail.com>, Heiko Strathmann <heiko.strathmann@gmail.com>, lambday <heavensdevil6909@gmail.com>, Roman Votyakov <votjakovr@gmail.com> | 08:04 |
-!- Cheng_ [cb8faf43@gateway/web/freenode/ip.203.143.175.67] has joined #shogun | 08:06 | |
-!- Cheng [80fa05f8@gateway/web/freenode/ip.128.250.5.248] has quit [Ping timeout: 250 seconds] | 08:06 | |
-!- sonne|osx [~sonne@82.113.106.234] has joined #shogun | 08:07 | |
-!- sonne|osx [~sonne@82.113.106.234] has quit [Ping timeout: 276 seconds] | 08:18 | |
-!- sonne|osx [~sonne@89.204.130.67] has joined #shogun | 08:21 | |
-!- sonne|osx [~sonne@89.204.130.67] has quit [Client Quit] | 08:26 | |
-!- foulwall [~zhengyang@114.255.40.22] has joined #shogun | 08:41 | |
-!- travis-ci [~travis-ci@ec2-54-211-149-66.compute-1.amazonaws.com] has joined #shogun | 08:42 | |
travis-ci | [travis-ci] it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/11671423 | 08:42 |
-!- travis-ci [~travis-ci@ec2-54-211-149-66.compute-1.amazonaws.com] has left #shogun [] | 08:42 | |
-!- Cheng_ [cb8faf43@gateway/web/freenode/ip.203.143.175.67] has left #shogun [] | 08:46 | |
-!- hushell [~hushell@c-98-232-178-161.hsd1.or.comcast.net] has joined #shogun | 08:47 | |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 09:02 | |
shogun-notifier- | shogun: Viktor Gal :develop * c2f889d / src/shogun/base/class_list.cpp.py: https://github.com/shogun-toolbox/shogun/commit/c2f889d79ac0f6ec2de1f79b3e9dc6d818fa93d4 | 09:02 |
shogun-notifier- | shogun: Add HAVE_NLOPT to class_list generator script for blacklisting | 09:02 |
shogun-buildbot | build #1797 of bsd1 - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/1797 blamelist: Viktor Gal <viktor.gal@maeth.com> | 09:08 |
shogun-buildbot | build #170 of FC19 - libshogun is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/FC19%20-%20libshogun/builds/170 | 09:20 |
-!- besser82 [~besser82@77-21-40-118-dynip.superkabel.de] has joined #shogun | 09:23 | |
-!- besser82 [~besser82@77-21-40-118-dynip.superkabel.de] has quit [Changing host] | 09:23 | |
-!- besser82 [~besser82@fedora/besser82] has joined #shogun | 09:23 | |
shogun-buildbot | build #1249 of rpm1 - libshogun is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/rpm1%20-%20libshogun/builds/1249 | 09:25 |
sonne|work | foulwall: OK? | 09:25 |
besser82 | G'Morning, Folks :D | 09:26 |
sonne|work | besser82: moin! | 09:26 |
besser82 | soone|work: Did you get my msg, from yesterday? | 09:26 |
besser82 | soone|work: About TJ's concern making LightSVM GPL'ed or sth.... | 09:26 |
sonne|work | besser82: yes | 09:27 |
besser82 | soone|work: Did hegive you any explanation, why? | 09:27 |
besser82 | sonne|work: Did hegive you any explanation, why? | 09:27 |
sonne|work | besser82: yeah | 09:27 |
sonne|work | not everyone likes that work can be used by others even commerically | 09:28 |
sonne|work | but besser82 write him an email maybe he changed his mind | 09:28 |
besser82 | sonne|work: I'll do, maybe AGPLv3+ might change his mind :) | 09:29 |
shogun-buildbot | build #135 of precise - libshogun is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/precise%20-%20libshogun/builds/135 | 09:29 |
sonne|work | besser82: no | 09:29 |
besser82 | sonne|work: hmm... | 09:29 |
besser82 | sonne|work: I'll write him; let's wait for the comeback.... | 09:30 |
shogun-buildbot | build #169 of FCRH - libshogun is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/FCRH%20-%20libshogun/builds/169 | 09:33 |
-!- gLando is now known as gsomix | 09:35 | |
-!- iglesiasg [~iglesias@dyna218-066.csc.kth.se] has joined #shogun | 09:38 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 09:38 | |
hushell | iglesiasg: hey | 09:48 |
@iglesiasg | hushell, hi! | 09:49 |
hushell | iglesiasg: how is going? | 09:49 |
@iglesiasg | hushell, just fine, for you? | 09:49 |
hushell | iglesiasg: could you help me merge the last PR? | 09:49 |
@iglesiasg | hushell, did you update the file loading after you submited the new data file in shogun-data? | 09:49 |
hushell | iglesiasg: just finished a workshop paper, I am working on my gsoc now :) | 09:49 |
hushell | iglesiasg: yep, you merged the .mat file for me | 09:50 |
shogun-buildbot | build #1529 of cyg1 - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/cyg1%20-%20libshogun/builds/1529 blamelist: Viktor Gal <viktor.gal@maeth.com> | 09:50 |
@iglesiasg | hushell, I mean in the notebook | 09:50 |
@iglesiasg | hushell, since the file changed, I guess the notebook needed to be updated as well | 09:50 |
hushell | iglesiasg: let me double check, but I remember it should be there | 09:50 |
hushell | iglesiasg: yeah, I used the correct file name | 09:51 |
@iglesiasg | hushell, oook then we should be ready to go | 09:52 |
hushell | iglesiasg: thanks! | 09:52 |
shogun-notifier- | shogun: hushell :develop * 9f3af69 / / (10 files): https://github.com/shogun-toolbox/shogun/commit/9f3af69b3624e10412d0efaa04bd40d13ea8ec19 | 09:52 |
shogun-notifier- | shogun: ipython notebook of factor graph model | 09:52 |
shogun-notifier- | shogun: Fernando Iglesias :develop * e3f0d4f / / (10 files): https://github.com/shogun-toolbox/shogun/commit/e3f0d4ff9cc81f6b89bbb4e4a33dad454b7bbe0e | 09:52 |
shogun-notifier- | shogun: Merge pull request #1607 from hushell/develop | 09:52 |
shogun-notifier- | shogun: | 09:52 |
shogun-notifier- | shogun: ipython notebook of factor graph model | 09:52 |
hushell | iglesiasg: I'll commit the SGD stuff before the deadline, still working on it :) | 09:53 |
@iglesiasg | hushell, all right, good luck! | 09:53 |
-!- new_lido [~walid@193.227.20.2] has joined #shogun | 09:53 | |
hushell | iglesiasg: thanks, I got an idea to rearrange a bit our SOSVM code, but let's talk about this later this week | 09:55 |
@iglesiasg | ok | 09:55 |
shogun-notifier- | shogun: van51 :develop * 069cb0e / doc/ipython-notebooks/classification/HashedDocDotFeatures.ipynb: https://github.com/shogun-toolbox/shogun/commit/069cb0ef89cc9101c542c5e6daced0c8ea492a50 | 10:00 |
shogun-notifier- | shogun: Added notebook for HashedDocDotFeaturs | 10:00 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * fff5bd5 / doc/ipython-notebooks/classification/HashedDocDotFeatures.ipynb: https://github.com/shogun-toolbox/shogun/commit/fff5bd5be23aa2368d72edc9f934f9ee2e63cb4a | 10:00 |
shogun-notifier- | shogun: Merge pull request #1660 from van51/feature/notebook | 10:00 |
shogun-notifier- | shogun: | 10:00 |
shogun-notifier- | shogun: Added notebook for HashedDocDotFeaturs | 10:00 |
sonne|work | gsomix: any news on protobuf? | 10:10 |
sonne|work | foulwall: ping? | 10:10 |
gsomix | sonne|work, yep. added set/get_matrix methods and types into Chunks.proto | 10:11 |
sonne|work | gsomix: ok so I guess vector / matrix are done then right? | 10:11 |
gsomix | sonne|work, writing macro and unit-tests | 10:13 |
gsomix | sonne|work, btw why we need uint64_t for lenght of vector? | 10:13 |
gsomix | *length | 10:13 |
sonne|work | gsomix: if so please send a PR for that and then we still have to do SGStringLists (which is easy - just the same as vector but multiple of them) and sparse | 10:13 |
sonne|work | gsomix: it might become bigger than uint32 at some point and protbuf will not waste bytes anyway so whenever you write uint32 there you could as well just use uint64 | 10:14 |
gsomix | sonne|work, ah, ok. | 10:14 |
sonne|work | gsomix: the only question remains if we want to represent sparse/strings as efficient as possible or as convenient as possible | 10:15 |
gsomix | sonne|work, so it means that sizes of messages should be uint64_t too, right? | 10:15 |
sonne|work | gsomix: I mean one could write out huge arrays with the values and then with the indices separately | 10:15 |
shogun-buildbot | build #1798 of bsd1 - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/1798 blamelist: hushell <dom343@gmail.com> | 10:15 |
sonne|work | gsomix: yeah why not | 10:15 |
sonne|work | (but 1MB * 2**32 is hard to reach I guess...) | 10:16 |
sonne|work | but doesn't hurt | 10:16 |
gsomix | ok | 10:16 |
sonne|work | gsomix: so any thoughts about sparse? | 10:17 |
sonne|work | gsomix: what do you prefer? writing out all nnz values in one go? | 10:17 |
sonne|work | so basically column compressed storage like? | 10:17 |
sonne|work | I guess that would be most efficient (space wise etc) for the file | 10:18 |
gsomix | sonne|work, << convenient as possible | <ShogunVersion> <SparseMatrix header> <SparseVector 1 header> <data messages> ... and so on | 10:18 |
gsomix | but I'm not sure | 10:18 |
sonne|work | and I also guess speed wise since you can write big chunks of 1 MB all the time | 10:19 |
sonne|work | one could do the same for strings | 10:19 |
sonne|work | I mean write out the huge block of strings (all strings concatenated) | 10:19 |
sonne|work | and the string lengths | 10:19 |
sonne|work | I guess we should do it this way since one might have say 100 million strings of small length | 10:20 |
gsomix | sonne|work, yep, I think so. | 10:20 |
sonne|work | and then the overhead would be drastic if not done otherwise | 10:20 |
sonne|work | same with sparse | 10:20 |
sonne|work | gsomix: so do strings first right after your next PR | 10:21 |
sonne|work | and hurry up man! | 10:21 |
gsomix | sonne|work, ok | 10:21 |
shogun-buildbot | build #1799 of bsd1 - libshogun is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/1799 | 10:22 |
-!- travis-ci [~travis-ci@ec2-54-243-23-104.compute-1.amazonaws.com] has joined #shogun | 10:22 | |
travis-ci | [travis-ci] it's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/11677793 | 10:22 |
-!- travis-ci [~travis-ci@ec2-54-243-23-104.compute-1.amazonaws.com] has left #shogun [] | 10:22 | |
gsomix | wiking, yo | 10:24 |
gsomix | sonne|work, btw what's better solution for message objects? keep it in stack and copy while returning from read_header()? or work with it on the heap+pointers? | 10:40 |
shogun-buildbot | build #1530 of cyg1 - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/cyg1%20-%20libshogun/builds/1530 blamelist: hushell <dom343@gmail.com>, Fernando Iglesias <fernando.iglesiasg@gmail.com> | 10:40 |
sonne|work | gsomix: I have no idea if they keep data on the stack - if they do then pointers! | 10:41 |
-!- van51 [~van51@athedsl-409794.home.otenet.gr] has joined #shogun | 10:45 | |
van51 | hello | 10:45 |
sonne|work | hey van51! | 10:45 |
sonne|work | I only had very minor comments - I guess you've seen. | 10:45 |
van51 | sonne|work: yeah I've seen them. Will make the changes now :) | 10:45 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 690e9de / README.cmake: https://github.com/shogun-toolbox/shogun/commit/690e9de1bd60b228f019c5db62ce5a8cfcff7e6e | 10:53 |
shogun-notifier- | shogun: mention how to bundle eigen3 | 10:53 |
shogun-buildbot | build #1531 of cyg1 - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/cyg1%20-%20libshogun/builds/1531 blamelist: Soeren Sonnenburg <sonne@debian.org>, van51 <vangelis_51@hotmail.com> | 11:13 |
-!- travis-ci [~travis-ci@ec2-54-211-149-66.compute-1.amazonaws.com] has joined #shogun | 11:17 | |
travis-ci | [travis-ci] it's Fernando Iglesias's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/11678897 | 11:17 |
-!- travis-ci [~travis-ci@ec2-54-211-149-66.compute-1.amazonaws.com] has left #shogun [] | 11:17 | |
-!- votjakovr [~votjakovr@host-46-241-3-209.bbcustomer.zsttk.net] has quit [Quit: Good bye!] | 11:17 | |
-!- votjakovr [~votjakovr@host-46-241-3-209.bbcustomer.zsttk.net] has joined #shogun | 11:22 | |
shogun-buildbot | build #1880 of deb3 - modular_interfaces is complete: Failure [failed test libshogun] Build details are at http://buildbot.shogun-toolbox.org/builders/deb3%20-%20modular_interfaces/builds/1880 blamelist: Soeren Sonnenburg <sonne@debian.org>, hushell <dom343@gmail.com>, Fernando Iglesias <fernando.iglesiasg@gmail.com>, van51 <vangelis_51@hotmail.com> | 11:27 |
-!- new_lido [~walid@193.227.20.2] has quit [Remote host closed the connection] | 11:34 | |
sonne|work | van51: done? | 11:41 |
van51 | sonne|work: yeah, pushed. For the setters I had to propagate them up to the converter that's why it touched more files | 11:42 |
shogun-buildbot | build #1532 of cyg1 - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/cyg1%20-%20libshogun/builds/1532 blamelist: Soeren Sonnenburg <sonne@debian.org> | 11:46 |
sonne|work | (van51 - I don't get notified at work so just tell me here) | 11:47 |
sonne|work | gsomix: you too ^ | 11:47 |
shogun-notifier- | shogun: van51 :develop * 58458df / / (6 files): https://github.com/shogun-toolbox/shogun/commit/58458df0de1c407dbf57d4dafa091defa9572d28 | 11:48 |
shogun-notifier- | shogun: Added full functionality of HashedDocDotFeatures to the streaming version and fixed a unit test file that failed at times | 11:48 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * a9b8552 / / (6 files): https://github.com/shogun-toolbox/shogun/commit/a9b8552b98213a16f67f8bca42c913fb4a5bb2fd | 11:48 |
shogun-notifier- | shogun: Merge pull request #1650 from van51/feature/fix_test | 11:48 |
shogun-notifier- | shogun: | 11:48 |
shogun-notifier- | shogun: StreamingHashedDocDotFeatures | 11:48 |
shogun-notifier- | shogun: van51 :develop * 08f892c / / (5 files): https://github.com/shogun-toolbox/shogun/commit/08f892c52085639128d901ff6dc9525f31ab3492 | 11:49 |
shogun-notifier- | shogun: Random kitchen sinks | 11:49 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * ec60496 / / (5 files): https://github.com/shogun-toolbox/shogun/commit/ec60496d26aebc2cd79e90949895ac4c891d2439 | 11:49 |
shogun-notifier- | shogun: Merge pull request #1606 from van51/feature/rks | 11:49 |
shogun-notifier- | shogun: | 11:49 |
shogun-notifier- | shogun: Random kitchen sink framework | 11:49 |
sonne|work | gsomix: any news? | 11:49 |
sonne|work | and foulwall ping again?! | 11:49 |
shogun-buildbot | build #36 of clang34 - static analysis is complete: Failure [failed analyse] Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20static%20analysis/builds/36 blamelist: Viktor Gal <viktor.gal@maeth.com> | 11:50 |
shogun-buildbot | build #21 of clang34 - undefined behaviour analysis is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20undefined%20behaviour%20analysis/builds/21 blamelist: Viktor Gal <viktor.gal@maeth.com> | 12:01 |
-!- travis-ci [~travis-ci@ec2-54-211-149-66.compute-1.amazonaws.com] has joined #shogun | 12:07 | |
travis-ci | [travis-ci] it's Soeren Sonnenburg's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/11679095 | 12:07 |
-!- travis-ci [~travis-ci@ec2-54-211-149-66.compute-1.amazonaws.com] has left #shogun [] | 12:07 | |
shogun-buildbot | build #25 of clang34 - thread analysis is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20thread%20analysis/builds/25 blamelist: Viktor Gal <viktor.gal@maeth.com> | 12:17 |
shogun-buildbot | build #30 of debian wheezy - memcheck is complete: Failure [failed memory check] Build details are at http://buildbot.shogun-toolbox.org/builders/debian%20wheezy%20-%20memcheck/builds/30 blamelist: Viktor Gal <viktor.gal@maeth.com> | 12:19 |
-!- HeikoS [~heiko@218.4.166.178.rev.vodafone.pt] has joined #shogun | 12:20 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 12:20 | |
shogun-notifier- | shogun: Roman Votyakov :develop * 24632fb / / (2 files): https://github.com/shogun-toolbox/shogun/commit/24632fbfc34fa88d3da1dd29d9cf96b51b9d8e62 | 12:30 |
shogun-notifier- | shogun: increase accuracy of gradient search and check parameters in unit test | 12:30 |
shogun-notifier- | shogun: Heiko Strathmann :develop * c30dbe2 / / (2 files): https://github.com/shogun-toolbox/shogun/commit/c30dbe2847c7d78b49a338be5a089eef5a1cdf2d | 12:30 |
shogun-notifier- | shogun: Merge pull request #1663 from votjakovr/feature/gp_refactoring | 12:30 |
shogun-notifier- | shogun: | 12:30 |
shogun-notifier- | shogun: Increase default accuracy of gradient search and check parameters in unit test | 12:30 |
-!- HeikoS [~heiko@218.4.166.178.rev.vodafone.pt] has quit [Quit: Leaving.] | 12:32 | |
shogun-buildbot | build #1533 of cyg1 - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/cyg1%20-%20libshogun/builds/1533 blamelist: Soeren Sonnenburg <sonne@debian.org>, van51 <vangelis_51@hotmail.com> | 12:53 |
-!- votjakovr [~votjakovr@host-46-241-3-209.bbcustomer.zsttk.net] has quit [Quit: Good bye!] | 13:11 | |
-!- zxtx_ [~zv@209-6-202-142.c3-0.smr-ubr2.sbo-smr.ma.cable.rcn.com] has quit [Ping timeout: 256 seconds] | 13:13 | |
shogun-buildbot | build #1881 of deb3 - modular_interfaces is complete: Failure [failed test python modular] Build details are at http://buildbot.shogun-toolbox.org/builders/deb3%20-%20modular_interfaces/builds/1881 blamelist: Soeren Sonnenburg <sonne@debian.org> | 13:25 |
shogun-buildbot | build #1882 of deb3 - modular_interfaces is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/deb3%20-%20modular_interfaces/builds/1882 | 13:35 |
-!- votjakovr [~votjakovr@host-46-241-3-209.bbcustomer.zsttk.net] has joined #shogun | 13:38 | |
-!- votjakovr [~votjakovr@host-46-241-3-209.bbcustomer.zsttk.net] has quit [Read error: Connection reset by peer] | 13:52 | |
gsomix | sonne|work, I was debugging. https://github.com/shogun-toolbox/shogun/pull/1651 | 13:54 |
sonne|work | gsomix: ok just one minor comment and then it is good | 13:56 |
sonne|work | wiking: how can we run protoc before compiling shogun? | 13:57 |
gsomix | sonne|work, wiking said that he need to write some hack for cmake. | 13:58 |
-!- votjakovr [~votjakovr@host-46-241-3-209.bbcustomer.zsttk.net] has joined #shogun | 13:58 | |
gsomix | sonne|work, but there is another problem. for now we cannot compile .cc files, but protoc produces code only in this way. | 13:58 |
sonne|work | gsomix: well we generate lib/versionstring.h already so it should be doable | 13:58 |
-!- lisitsyn1 [~lisitsin@mxs.kg.ru] has joined #shogun | 13:59 | |
besser82 | sonne|work: I can implement a way on refactoring CMake-buildsys :) | 13:59 |
shogun-buildbot | build #1534 of cyg1 - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/cyg1%20-%20libshogun/builds/1534 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Roman Votyakov <votjakovr@gmail.com> | 13:59 |
besser82 | sonne|work: will be finished within next ~2-3 days | 14:00 |
sonne|work | besser82: not sure what you are saying but any help welcome :) | 14:00 |
besser82 | sonne|work: I currently refactoring the CMake-buildsys in agreement with wiking, to get out all glitches and/or quick'n'dirty workarounds ;) | 14:01 |
sonne|work | ahh nice | 14:01 |
gsomix | sonne|work, ok. I need cup of tea and then will do string lists | 14:02 |
besser82 | sonne|work: so everything will be run out of src tree and lots of other stuff to come ;) | 14:02 |
besser82 | sonne|work: even with testsuite being able to be run from build-tree ;) | 14:02 |
sonne|work | gsomix: please fix the PR and then comment the PROTOBUF detection and let me merge it | 14:03 |
sonne|work | then continue | 14:03 |
sonne|work | besser82: ahh nice highly needed | 14:03 |
sonne|work | besser82: even more important currently is that one cannot build shogun when it is already installed in /usr/local due to include clashes | 14:04 |
besser82 | sonne|work: np. will be fixed, too ;) | 14:04 |
sonne|work | so include path order needs to be changed | 14:04 |
sonne|work | besser82: ahh ok then just do whatever you do :) | 14:04 |
besser82 | sonne|work, wiking: do you agree with me to restructure source-tree as well? Like placing headers in seperate dir and suchss | 14:05 |
besser82 | s/ss/?/g | 14:05 |
sonne|work | besser82: errm you mean shogun/src ? | 14:06 |
besser82 | sonne|work, wiking: yes | 14:06 |
sonne|work | what is wrong / what would you do? | 14:06 |
besser82 | sonne|work: nothing really wrong, but would be abit "?bersichtlicher" ;) | 14:07 |
sonne|work | well make an example | 14:07 |
besser82 | sonne|work: now it is src/shogun/*/*.{h,c,cpp,py,...} | 14:08 |
sonne|work | yes | 14:08 |
sonne|work | and cxx once swig is done | 14:09 |
sonne|work | and .cc for protobuf later too | 14:09 |
besser82 | sonne|work: I think having it like src/include, src/shogun and so on | 14:09 |
besser82 | sonne|work: swig and protobuf will get generated out of tree ;) | 14:09 |
besser82 | sonne|work: like build/$cloned_dir_structure/... | 14:10 |
sonne|work | sounds very reasonable | 14:10 |
sonne|work | but I cannot speak for all | 14:10 |
besser82 | sonne|work: who else to ask? | 14:11 |
sonne|work | iglesiasg, lisitsyn, wiking, van51 - any thoughts on besser82's suggestion? | 14:11 |
@iglesiasg | let me read | 14:11 |
sonne|work | besser82: well HeikoS and the ones above :) | 14:11 |
@iglesiasg | besser82, so headers in a separate soource tree? | 14:11 |
@iglesiasg | I think I have to look up "?bersichtlicher" :D | 14:12 |
besser82 | iglesiasg: yes, and probably *.py as well | 14:12 |
besser82 | iglesiasg: ?bersichtlicher == clear, more semantic ;) | 14:13 |
@iglesiasg | besser82, hehe yeah just saw | 14:13 |
@iglesiasg | besser82, so it is basically a matter of taste, right? | 14:13 |
@iglesiasg | no technical advantages/disadvantages? | 14:13 |
besser82 | iglesiasg: a matter of taste, common-sense and good pratice ;) | 14:13 |
besser82 | iglesiasg: mashing up all in one dir or so was from "old" autocrap" times ;) | 14:14 |
@iglesiasg | besser82, hehe | 14:14 |
@iglesiasg | for me it is fine, sure | 14:14 |
besser82 | iglesiasg: kk, just the rest's agreement required, then ;) | 14:15 |
@iglesiasg | I think I have seen this headers/implementation files separation in PCL before | 14:15 |
@iglesiasg | the first second I was like wtf!?! hehe | 14:15 |
besser82 | iglesiasg: and some other projects as well ;) might be a bit confusing on the first look :D | 14:16 |
@iglesiasg | :) | 14:16 |
besser82 | iglesiasg: main goal is to have clear seperation between sources and build-tree, headers and impl, different stuff from impl | 14:17 |
sonne|work | besser82: what needs attention here is that we want at some point follow the d pointer approach | 14:17 |
sonne|work | besser82: i.e. only have interfaces .h files exposed to the outside | 14:17 |
sonne|work | and encapsulate the rest in Impl classes | 14:18 |
sonne|work | so one would just have to copy headers over | 14:18 |
@iglesiasg | sonne|work, let's check how they do it in QT then | 14:18 |
sonne|work | currently headers are both interface and whatever | 14:18 |
besser82 | sonne|work: so you want to have distiction between interf-haeders and impl-headers, right? | 14:19 |
sonne|work | yes | 14:19 |
sonne|work | I mean if you restructure then lets not do that again in a few months | 14:19 |
besser82 | sonne|work: would be really easy with that split dir approach ;) | 14:20 |
@iglesiasg | so QT people have in the same dirs cpp, and h files | 14:20 |
sonne|work | well how? | 14:20 |
besser82 | sonne|work: let's split-up the include-dir to interface and impl ;) | 14:20 |
@iglesiasg | but maybe I guess it should not be a limitation to separate them in order to use d pointer | 14:20 |
besser82 | iglesiasg, sonne|work: let's split-up the include-dir to interface and impl ;) | 14:21 |
lisitsyn1 | yes I like splitting | 14:21 |
lisitsyn1 | besser82: sonne|work: ^ | 14:21 |
besser82 | iglesiasg, sonne|work: like include/implementation include/public | 14:21 |
sonne|work | lisitsyn1 prolly A.vim won't work any longer | 14:21 |
* besser82 lisitsyn1 +1 :D | 14:21 | |
lisitsyn1 | sonne|work: yeah true | 14:22 |
besser82 | sonne|work, lisitsyn1: you don't use eclipse ??? XD | 14:23 |
sonne|work | iglesiasg, besser82 actually the d ptr stuff shouldn't need includes at all or? | 14:23 |
shogun-buildbot | build #37 of clang34 - static analysis is complete: Failure [failed analyse] Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20static%20analysis/builds/37 blamelist: van51 <vangelis_51@hotmail.com> | 14:23 |
sonne|work | besser82: nope vim | 14:23 |
sonne|work | I mean non public includes | 14:23 |
besser82 | sonne|work: afaik, no | 14:23 |
lisitsyn1 | besser82: I don't | 14:24 |
sonne|work | besser82: I am failing to do this properly with C++ | 14:26 |
besser82 | sonne|work: what dpointer, seperated includes or eclipse??? | 14:26 |
@iglesiasg | besser82, http://qt-project.org/wiki/Dpointer | 14:27 |
@iglesiasg | besser82, and this one http://en.wikipedia.org/wiki/Opaque_pointer | 14:27 |
besser82 | sonne|work: So dpointers should be pretty straight forward with c++ | 14:29 |
gsomix | sonne|work, done. https://github.com/shogun-toolbox/shogun/pull/1651 | 14:29 |
sonne|work | gsomix: github down? | 14:30 |
gsomix | sonne|work, nope | 14:30 |
sonne|work | unreachable here | 14:30 |
besser82 | sonne|work: let me finish-off that cmake and then let's get those dpointers done :D | 14:30 |
sonne|work | besser82: just do :) | 14:31 |
shogun-buildbot | build #31 of debian wheezy - memcheck is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/debian%20wheezy%20-%20memcheck/builds/31 | 14:31 |
besser82 | but if there's anybody out here with big affinity or interest in RHEL/CentOS, I might need some help on backporting lots of stuff to them | 14:33 |
besser82 | like python2.6, eigen3, even more ;) | 14:34 |
sonne|work | besser82: there probably is but not here I am afraid | 14:36 |
besser82 | sonne|work: I'm afraid, too :( But I just wanted to ask here, since it's perfectly about BRs of SHOGUN ;) | 14:37 |
lisitsyn1 | yeah d-pointers are easy | 14:37 |
sonne|work | BRs ? | 14:38 |
lisitsyn1 | struct Impl; | 14:38 |
lisitsyn1 | Impl* impl; | 14:38 |
lisitsyn1 | that's it :D | 14:38 |
besser82 | sonne|work: BuildRequires | 14:38 |
besser82 | sonne|work: like build-deps in Debian ;) | 14:38 |
sonne|work | lisitsyn1: yeah but the difficulty is to do the proper interface - so not at all easy! | 14:38 |
lisitsyn1 | sonne|work: what inteface? | 14:38 |
besser82 | lisitsyn1: to the hidden impl ;) | 14:39 |
sonne|work | yeah which functions do you expose and which not | 14:39 |
lisitsyn1 | it is up to you, what is proper here? | 14:39 |
sonne|work | this decision! | 14:39 |
lisitsyn1 | you don't expose anything | 14:39 |
sonne|work | yeah difficult to decide | 14:39 |
besser82 | lisitsyn1: you expose public interfaces | 14:39 |
besser82 | lisitsyn1: which should be kept binary compatible | 14:40 |
lisitsyn1 | class PublicObject | 14:40 |
lisitsyn1 | { | 14:40 |
lisitsyn1 | public: void doStuff(); | 14:40 |
lisitsyn1 | private: struct Impl; Impl* impl; | 14:40 |
lisitsyn1 | } | 14:40 |
besser82 | lisitsyn1: that's what dpointer are about ;) | 14:40 |
lisitsyn1 | what to expose here? | 14:40 |
sonne|work | van51: btw nice ipython notebook! | 14:41 |
lisitsyn1 | besser82: are you talking about setting that Impl with some public interface to stone? | 14:43 |
lisitsyn1 | I don't care about binary compatibility to be honest hehe | 14:44 |
lisitsyn1 | I don't see we slow down API changes | 14:45 |
van51 | sonne|work: thanks! let me know if you have any suggestions about it | 14:46 |
sonne|work | van51: Well only thing that is sub-optimal is that you didnt' re-run the tests for spectrum etc in that comparison | 14:47 |
sonne|work | apart from that all good | 14:47 |
van51 | sonne|work: hmm I can run them then and update it accordingly | 14:48 |
@iglesiasg | van51, I am reading it atm :) | 14:48 |
sonne|work | van51: maybe this is sth for your kitchen sinks http://www.robots.ox.ac.uk/~vedaldi/assets/pubs/vedaldi11efficient.pdf | 14:49 |
sonne|work | we have that as homegenous kernel map currently in shogun | 14:49 |
sonne|work | but not as dot features | 14:49 |
van51 | iglesiasg: cool :) if you also have any suggestions feel free to tell me! | 14:50 |
@iglesiasg | van51, one question! | 14:50 |
van51 | sonne|work: I will take a look | 14:50 |
van51 | iglesiasg: shoot | 14:50 |
@iglesiasg | van51, "First of all, the dimensionality of the feature space (or else, the number of distinct tokens in the collection) is usually not known beforehand and requires a pass over the entire dataset to be calculated. This can make the creation of the document-term matrix trickier." | 14:50 |
@iglesiasg | van51, why does it get trickier if you have to do a first pass to count the number of distinct tokens? | 14:51 |
sonne|work | iglesiasg: online learning? | 14:51 |
sonne|work | iglesiasg: you don't know the number of dims beforehand | 14:51 |
@iglesiasg | true, it is an issue in online learning | 14:51 |
sonne|work | so what do you do then | 14:51 |
@iglesiasg | I didn't consider it | 14:52 |
@iglesiasg | van51, so a shy suggestion from my part is to say that ^ becomes an issue in online learning :) | 14:53 |
van51 | iglesiasg: yes I could elaborate a bit more there | 14:55 |
sonne|work | iglesiasg: even in the standard case you would have to compute all e.g. n-grams to determine the biggest index | 14:56 |
@iglesiasg | sonne|work, but that is something you would need to do afterwards anyway, right? | 14:57 |
sonne|work | iglesiasg: well you do it while training only... | 14:58 |
@iglesiasg | sonne|work, sure. I am talking about training | 14:58 |
sonne|work | so you don't have these all available explictly | 14:58 |
@iglesiasg | van51, cool notebook indeed! It's nice to know more about your project | 15:02 |
van51 | iglesiasg: I'm glad that you like it | 15:02 |
van51 | all the other notebooks I 've seen looked very detailed and well-worked that I was worried that mine was kind of sloppy | 15:03 |
shogun-buildbot | build #38 of clang34 - static analysis is complete: Failure [failed analyse] Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20static%20analysis/builds/38 blamelist: Soeren Sonnenburg <sonne@debian.org>, van51 <vangelis_51@hotmail.com> | 15:03 |
@iglesiasg | all right, lunch time. See you! | 15:06 |
-!- iglesiasg [~iglesias@dyna218-066.csc.kth.se] has quit [Quit: Ex-Chat] | 15:06 | |
shogun-buildbot | build #22 of clang34 - undefined behaviour analysis is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20undefined%20behaviour%20analysis/builds/22 blamelist: Soeren Sonnenburg <sonne@debian.org>, van51 <vangelis_51@hotmail.com> | 15:13 |
-!- travis-ci [~travis-ci@ec2-107-22-23-41.compute-1.amazonaws.com] has joined #shogun | 15:17 | |
travis-ci | [travis-ci] it's Soeren Sonnenburg's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/11680511 | 15:17 |
-!- travis-ci [~travis-ci@ec2-107-22-23-41.compute-1.amazonaws.com] has left #shogun [] | 15:17 | |
-!- HeikoS [~heiko@218.4.166.178.rev.vodafone.pt] has joined #shogun | 15:22 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 15:23 | |
shogun-buildbot | build #26 of clang34 - thread analysis is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20thread%20analysis/builds/26 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Soeren Sonnenburg <sonne@debian.org>, van51 <vangelis_51@hotmail.com>, Roman Votyakov <votjakovr@gmail.com> | 15:28 |
-!- HeikoS [~heiko@218.4.166.178.rev.vodafone.pt] has quit [Quit: Leaving.] | 15:30 | |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 15:30 | |
-!- HeikoS [~heiko@218.4.166.178.rev.vodafone.pt] has joined #shogun | 15:31 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 15:31 | |
@HeikoS | votjakovr: hey! | 15:41 |
@HeikoS | votjakovr: thanks for the latest patch, any thoughts about the surface plot of the ML? | 15:42 |
@HeikoS | votjakovr: also, I wanted to hear about the notebook | 15:43 |
@HeikoS | did you see the draft? | 15:43 |
sonne|work | besser82: there is HeikoS! | 15:43 |
@HeikoS | sonne|work: ah, what did I do wrong? :) | 15:43 |
besser82 | HeikoS: nothing ;) | 15:43 |
@HeikoS | besser82: hi then :) | 15:43 |
besser82 | HeikoS: You'rethe last one to vote ;) | 15:43 |
@HeikoS | vote? | 15:43 |
@HeikoS | GPU based eigen3 backends :D | 15:44 |
@HeikoS | ?? | 15:44 |
besser82 | HeikoS: Let's make that german ;) | 15:44 |
@HeikoS | distributed matrix matrix multiplication :D ? :) | 15:44 |
votjakovr | HeikoS: hi! I haven't think about surface plot yet :) Now i'm working on hiv example | 15:46 |
@HeikoS | votjakovr: okay, dont worry aobut the surface plot for now, maybe later | 15:48 |
votjakovr | HeikoS: how is it better to solve the problem? classify R5X4 vs all? | 15:48 |
@HeikoS | votjakovr: okay cool, do things work? | 15:48 |
@HeikoS | votjakovr: did you have a look into the paper? | 15:48 |
votjakovr | HeikoS: very briefly | 15:48 |
@HeikoS | votjakovr: I forgot what I exactly did back then, but got very good performance | 15:49 |
@HeikoS | votjakovr: yeah I think thats a sensible idea to do | 15:51 |
votjakovr | HeikoS: just for sure: we get protein sequences and usage of coreceptor | 15:52 |
@HeikoS | votjakovr: yeah you just get two sets of sequences for the alphabet of aminoacids (20 letters) | 15:52 |
@HeikoS | votjakovr: there might be some letters in there that correspond do "dont know" or so, and shogun might complain about that, but its not very important for distant-segments string kernel, only for the specturm kernel we have in shogun | 15:53 |
besser82 | sonne|work: Heiko says +1, too :D | 15:54 |
votjakovr | HeikoS: hmm i've downloaded the data and protein's string length are different | 15:56 |
votjakovr | HeikoS: ~ 33-37 characters | 15:56 |
@HeikoS | votjakovr: yes, thats the point of string kernels | 15:56 |
@HeikoS | they can handle different lengths | 15:56 |
@HeikoS | votjakovr: magic, isnt it ? :) | 15:57 |
votjakovr | HeikoS: ok cool :) | 15:57 |
votjakovr | HeikoS: so we decided to classify R5X4 vs R5 and X4, right? | 15:58 |
@HeikoS | votjakovr: the kernel has two parameters | 15:58 |
@HeikoS | votjakovr: which you cannot choose via gradient descent | 15:58 |
@HeikoS | they are discrete | 15:58 |
votjakovr | HeikoS: yep i see | 15:58 |
@HeikoS | votjakovr: so I just realised we at some point need something to do a grid-search using the marginal likelihood ;) | 15:58 |
votjakovr | HeikoS: ah :) | 15:59 |
@HeikoS | votjakovr: but that also later | 15:59 |
@HeikoS | the string example is just to demonstrate how cool we are ;) | 15:59 |
votjakovr | HeikoS: huh | 15:59 |
votjakovr | HeikoS: BTW thank you, for the cool notebook draft :) | 16:04 |
@HeikoS | votjakovr: no worries, its just a very basic skeleton yet. I will do some parts for toy GP classification now if you dont mind? | 16:04 |
votjakovr | HeikoS: ok, thanks | 16:05 |
shogun-buildbot | build #39 of clang34 - static analysis is complete: Failure [failed analyse] Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20static%20analysis/builds/39 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Roman Votyakov <votjakovr@gmail.com> | 16:10 |
shogun-buildbot | build #23 of clang34 - undefined behaviour analysis is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20undefined%20behaviour%20analysis/builds/23 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Roman Votyakov <votjakovr@gmail.com> | 16:19 |
besser82 | lisitsyn1: https://koji.fedoraproject.org/koji/packageinfo?packageID=17051 :D | 16:24 |
besser82 | lisitsyn1: in repos for tomorrow :D | 16:25 |
votjakovr | HeikoS: hmm... unfortunately we have his: http://github.com/shogun-toolbox/shogun/blob/develop/src/shogun/machine/gp/InferenceMethod.cpp#L304 | 16:37 |
-!- zxtx [~zv@209-6-202-142.c3-0.smr-ubr2.sbo-smr.ma.cable.rcn.com] has joined #shogun | 16:38 | |
@HeikoS | votjakovr: ah shit | 16:38 |
@HeikoS | votjakovr: what do you think, how much effort is it to make things work for other types of features? | 16:38 |
votjakovr | HeikoS: hmm... how could we this say for string features? | 16:42 |
@HeikoS | votjakovr: what is m_feat used for? | 16:42 |
-!- van51 [~van51@athedsl-409794.home.otenet.gr] has quit [Read error: No route to host] | 16:43 | |
votjakovr | HeikoS: this is X - matrix, which contains features | 16:43 |
@HeikoS | but it is the dot product wiht itself right? | 16:44 |
votjakovr | HeikoS: ah, ok | 16:45 |
votjakovr | HeikoS: it is used for mean | 16:46 |
-!- van51 [~van51@athedsl-409794.home.otenet.gr] has joined #shogun | 16:46 | |
votjakovr | HeikoS: to compute values of mean function | 16:46 |
votjakovr | HeikoS: and derivatives | 16:47 |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 16:47 | |
shogun-notifier- | shogun: Heiko Strathmann :gp_notebook * 8051c74 / doc/ipython-notebooks/gaussian_process/gaussian_process_classification.ipynb: https://github.com/shogun-toolbox/shogun/commit/8051c74824634677b29b182e0ff7f2a56003347a | 16:47 |
shogun-notifier- | shogun: some updates on binary classification | 16:47 |
@HeikoS | votjakovr: the mean function in shogun is limited to R^n right? | 16:48 |
@HeikoS | votjakovr: ah thats annoying, I totally missed that | 16:48 |
votjakovr | HeikoS: currently yes | 16:48 |
@HeikoS | votjakovr: in fact the mean function should be defined as m(x) : Y^n -> R | 16:48 |
@HeikoS | shit | 16:49 |
@HeikoS | its like the kernel/covariance, defined on features | 16:49 |
@HeikoS | votjakovr: ok, we cannot change that now | 16:49 |
@HeikoS | have to postpone | 16:49 |
@HeikoS | votjakovr: sorry for having missed that :( | 16:49 |
@HeikoS | votjakovr: I mean it is possible, but we will have to change a few internals for that | 16:49 |
votjakovr | HeikoS: ah, me too | 16:49 |
@HeikoS | votjakovr: ok then, we will have to skip the string GP | 16:49 |
@HeikoS | and rather go for some other std datasets | 16:49 |
@HeikoS | votjakovr: I suggest you do the UCI glass dataset | 16:50 |
@HeikoS | window-glass vs non-window glass | 16:50 |
@HeikoS | in fact, I should have code for that somewhere, let me search | 16:50 |
@HeikoS | votjakovr: https://github.com/karlnapf/kameleon-mcmc/blob/master/main/gp/GPData.py | 16:51 |
@HeikoS | there I load the glass data, its the file from | 16:51 |
@HeikoS | votjakovr: http://archive.ics.uci.edu/ml/datasets/Glass+Identification | 16:51 |
@HeikoS | votjakovr: and https://github.com/karlnapf/kameleon-mcmc/blob/master/main/gp/scripts/glass_gaussian_ard.py loads the data and preprocesses it | 16:52 |
@HeikoS | votjakovr: maybe start with that first | 16:53 |
@HeikoS | votjakovr: btw I just updated the notebook, so make sure to rebase from time to time to avoid clashes | 16:53 |
votjakovr | HeikoS: oh, thanks | 16:53 |
@HeikoS | votjakovr: I have to go now, will be back later tonight, might be after the deadline | 16:53 |
@HeikoS | votjakovr: but dont worry, will still merge things then :) | 16:54 |
votjakovr | HeikoS: ok :) | 16:54 |
@HeikoS | votjakovr: but try to finish this glass dataset at least, and if you find time, go for some regression data | 16:54 |
@HeikoS | votjakovr: if things are not completely polished by tonight, we can still improve things before the release. Though it would be good if you added stuff today. | 16:55 |
@HeikoS | votjakovr: ok, see you later! | 16:55 |
votjakovr | HeikoS: see you | 16:55 |
-!- HeikoS [~heiko@218.4.166.178.rev.vodafone.pt] has quit [Quit: Leaving.] | 16:56 | |
-!- travis-ci [~travis-ci@ec2-54-226-199-227.compute-1.amazonaws.com] has joined #shogun | 17:04 | |
travis-ci | [travis-ci] it's Soeren Sonnenburg's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/11682141 | 17:04 |
-!- travis-ci [~travis-ci@ec2-54-226-199-227.compute-1.amazonaws.com] has left #shogun [] | 17:04 | |
lisitsyn1 | besser82: cool thanks! | 17:13 |
besser82 | lisitsyn1: np. :D | 17:14 |
-!- sonne|osx [~sonne@89.204.139.220] has joined #shogun | 17:20 | |
shogun-notifier- | shogun: Evgeniy Andreev :develop * 478fe75 / / (8 files): https://github.com/shogun-toolbox/shogun/commit/478fe75987dc676dd9733d3e52f582cdf70dc7d2 | 17:25 |
shogun-notifier- | shogun: add ProtobufFile | 17:25 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * fb05e45 / / (8 files): https://github.com/shogun-toolbox/shogun/commit/fb05e459d2df05cc3c5551794c662945a1932b8a | 17:25 |
shogun-notifier- | shogun: Merge pull request #1651 from gsomix/feature/protobuf | 17:25 |
shogun-notifier- | shogun: | 17:25 |
shogun-notifier- | shogun: Protobuf based binary fileformat for reading/writing vectors/matrices with shogun | 17:25 |
shogun-notifier- | shogun: Fernando Iglesias :develop * f3a668d / src/shogun/evaluation/CrossValidationOutput.h: https://github.com/shogun-toolbox/shogun/commit/f3a668dfb28d71564467bbe863245467959cacae | 17:26 |
shogun-notifier- | shogun: Fix doc typo | 17:26 |
shogun-notifier- | shogun: Fernando Iglesias :develop * 5445172 / doc/ipython-notebooks/metric/LMNN.ipynb: https://github.com/shogun-toolbox/shogun/commit/544517222e45c5a1d6c553a54bd70efabde113d4 | 17:26 |
shogun-notifier- | shogun: Add multiclass classification illustration to LMNN notebook | 17:26 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 8be7bb0 / / (2 files): https://github.com/shogun-toolbox/shogun/commit/8be7bb0e6f105dfd4c84257902b082c32eb2c71d | 17:26 |
shogun-notifier- | shogun: Merge pull request #1657 from iglesias/feature/lmnn | 17:26 |
shogun-notifier- | shogun: | 17:26 |
shogun-notifier- | shogun: Feature/lmnn | 17:26 |
shogun-buildbot | build #2265 of deb1 - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun/builds/2265 blamelist: Evgeniy Andreev <gsomix@gmail.com>, Fernando Iglesias <fernando.iglesiasg@gmail.com> | 17:29 |
shogun-buildbot | build #2266 of deb1 - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun/builds/2266 blamelist: Soeren Sonnenburg <sonne@debian.org> | 17:30 |
gsomix | sonne|osx, sonne|osx sth wrong. Protobuf is in classlist. | 17:31 |
sonne|osx | gsomix: ifdef HAVE_PROTOBUF? | 17:31 |
gsomix | sonne|osx, yep. in *.h and *.cpp both. | 17:32 |
sonne|osx | gsomix: errm cannot see it | 17:33 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 345c989 / src/shogun/base/class_list.cpp.py: https://github.com/shogun-toolbox/shogun/commit/345c989498417cd3e14768049410d4744548d89c | 17:34 |
shogun-notifier- | shogun: add HAVE_PROTOBUF to classlist .py | 17:34 |
sonne|osx | that should do it | 17:34 |
sonne|osx | gsomix: that is a bit tricky - the .cpp is autogenerated so that might explain it | 17:35 |
gsomix | sonne|osx, ok. I'm finishing string_list stuff. | 17:36 |
sonne|osx | gsomix: send a PR as soon as possible for review | 17:38 |
gsomix | sonne|osx, ok, as soon as I get working unit-test. | 17:39 |
sonne|osx | gsomix: ok | 17:39 |
sonne|osx | gtg now | 17:39 |
sonne|osx | cu | 17:39 |
-!- sonne|osx [~sonne@89.204.139.220] has quit [Quit: sonne|osx] | 17:39 | |
shogun-buildbot | build #2267 of deb1 - libshogun is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun/builds/2267 | 17:40 |
-!- van51 [~van51@athedsl-409794.home.otenet.gr] has quit [Quit: Leaving.] | 17:43 | |
shogun-buildbot | build #1803 of bsd1 - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/1803 blamelist: Soeren Sonnenburg <sonne@debian.org> | 17:47 |
shogun-buildbot | build #2264 of deb1 - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun/builds/2264 blamelist: Fernando Iglesias <fernando.iglesiasg@gmail.com> | 17:48 |
-!- travis-ci [~travis-ci@ec2-107-22-23-41.compute-1.amazonaws.com] has joined #shogun | 17:55 | |
travis-ci | [travis-ci] it's Soeren Sonnenburg's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/11682164 | 17:55 |
-!- travis-ci [~travis-ci@ec2-107-22-23-41.compute-1.amazonaws.com] has left #shogun [] | 17:55 | |
shogun-buildbot | build #1535 of cyg1 - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/cyg1%20-%20libshogun/builds/1535 blamelist: Soeren Sonnenburg <sonne@debian.org> | 18:12 |
-!- foulwall [~zhengyang@114.255.40.22] has quit [Ping timeout: 252 seconds] | 18:25 | |
-!- lisitsyn1 [~lisitsin@mxs.kg.ru] has left #shogun [] | 18:28 | |
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has joined #shogun | 18:38 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 18:38 | |
-!- besser82 [~besser82@fedora/besser82] has quit [Quit: freedom, friends, features, first ---> fedoraproject.org] | 18:45 | |
-!- besser82 [~besser82@fedora/besser82] has joined #shogun | 18:47 | |
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has quit [Quit: Ex-Chat] | 18:53 | |
-!- travis-ci [~travis-ci@ec2-54-226-199-227.compute-1.amazonaws.com] has joined #shogun | 19:00 | |
travis-ci | [travis-ci] it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/11683455 | 19:00 |
-!- travis-ci [~travis-ci@ec2-54-226-199-227.compute-1.amazonaws.com] has left #shogun [] | 19:00 | |
-!- van51 [~van51@athedsl-409794.home.otenet.gr] has joined #shogun | 19:04 | |
-!- sonne|osx [~sonne@g225081034.adsl.alicedsl.de] has joined #shogun | 19:38 | |
-!- sonimperator is now known as sonney2k | 19:40 | |
-!- zxtx [~zv@209-6-202-142.c3-0.smr-ubr2.sbo-smr.ma.cable.rcn.com] has quit [Ping timeout: 264 seconds] | 19:42 | |
shogun-buildbot | build #1884 of deb3 - modular_interfaces is complete: Failure [failed test python modular] Build details are at http://buildbot.shogun-toolbox.org/builders/deb3%20-%20modular_interfaces/builds/1884 blamelist: Soeren Sonnenburg <sonne@debian.org> | 19:55 |
gsomix | sonney2k, sonne|osx https://github.com/shogun-toolbox/shogun/pull/1667 | 20:18 |
wiking | gsomix: u were looking for me? | 20:21 |
gsomix | wiking, yep. we need enable protoc and *.cc. :) | 20:21 |
gsomix | some of code for protobuf is already in develop branch. | 20:22 |
wiking | ok | 20:22 |
wiking | i'll give a look at it | 20:22 |
gsomix | 18:27 UTC. Epic fail. | 20:28 |
gsomix | time to change music in player from Om to Sunn O))) | 20:30 |
van51 | sonne|osx: can you also have a look here : https://github.com/shogun-toolbox/shogun/pull/1666 ? | 20:32 |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 20:34 | |
@sonney2k | gsomix, well done | 20:36 |
gsomix | sonney2k, ok. fixing types | 20:36 |
@sonney2k | gsomix, now all that is left after the minor fixes is sparse | 20:36 |
gsomix | and sparse next, yes. it will be easy and very similar to string lists | 20:37 |
@sonney2k | gsomix, lets just now discuss sparse matrices | 20:37 |
@sonney2k | I think it is best if you write it out in column compressed storage http://netlib.org/linalg/html_templates/node92.html#SECTION00931200000000000000 | 20:37 |
@sonney2k | which means you have to write out 3 arrays | 20:38 |
@sonney2k | gsomix, around and understood? | 20:40 |
gsomix | sonney2k, yep | 20:40 |
@sonney2k | it might be a bit confusing in the beginning... | 20:40 |
gsomix | reading | 20:40 |
@sonney2k | gsomix, basically all the nonzero entries are in the val array | 20:40 |
@sonney2k | gsomix, these are the things we have in the SGSparseVectorEntry.entry's | 20:41 |
@sonney2k | gsomix, actually you could do it indeed the same way like strings | 20:42 |
@sonney2k | gsomix, you just have to write the values in the entry (the way you write strings) and in addition the values feat_index (also the same way you write strings) | 20:43 |
@sonney2k | and then an array with the lengths of the sparse vectors (same as length of strings) | 20:43 |
gsomix | sonney2k, aha, I see. | 20:44 |
@sonney2k | gsomix, this is then almost the CCS format | 20:45 |
@sonney2k | gsomix, basically val == array of entries | 20:45 |
@sonney2k | row_ind == array of feat_index | 20:45 |
@sonney2k | col_ptr is different | 20:45 |
@sonney2k | it is cumsum(sparse vector lengths) | 20:46 |
@sonney2k | gsomix, does that make sense / is it clear? | 20:46 |
gsomix | sonney2k, but there is one question. should I store val and row_ind in series? or I can write pairs? | 20:46 |
@sonney2k | or Qs? | 20:46 |
@sonney2k | gsomix, write it in series | 20:46 |
@sonney2k | no pairs (slow!) | 20:46 |
gsomix | I mean one data message is <<part of val array> <part of row_ind array>> | 20:47 |
gsomix | or <part of val> <part of val> ... <part of row_ind> <part of row_ind> ... and so on | 20:47 |
@sonney2k | gsomix, like this | 20:48 |
@sonney2k | repeated int32 feat_indices; | 20:49 |
@sonney2k | repeated (whatever chunk type) values; | 20:49 |
@sonney2k | van51, looks nice - but give some example with a dataset we have in data for people to try out | 20:49 |
@sonney2k | van51, maybe even create a data symlink in that application dir | 20:50 |
@sonney2k | van51, and then have it ready | 20:50 |
gsomix | sonney2k, in one message? so I need add new types in Chunks, right? | 20:50 |
gsomix | ah, ok, it's clear | 20:52 |
-!- HeikoS [~heiko@218.4.166.178.rev.vodafone.pt] has joined #shogun | 20:53 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 20:53 | |
@sonney2k | gsomix, no you can also do it alternating in two messages | 20:55 |
@sonney2k | gsomix, I mean first indices then data | 20:56 |
votjakovr | HeikoS: why can't i do something like: features.add_preprocessor(new CPruneVarSubMean()) and features.apply_preprocessor() ? | 21:04 |
@HeikoS | votjakovr: what happens? I never used preprocessors that much ... | 21:04 |
votjakovr | HeikoS: it shows an exception | 21:05 |
@HeikoS | which is? | 21:07 |
@HeikoS | votjakovr: using preprocessors would be nice, though some are not really tested and may show weird behaviour :) | 21:08 |
@HeikoS | votjakovr: you might have to debug .. :( | 21:08 |
votjakovr | HeikoS: it tells [ERROR] In file /home/votjak/Projects/shogun/src/shogun/preprocessor/PruneVarSubMean.cpp line 127: assertion m_initialized failed in virtual SGMatrix<float64_t> shogun::CPruneVarSubMean::apply_to_feature_matrix(shogun::CFeatures *) file /home/votjak/Projects/shogun/src/shogun/preprocessor/PruneVarSubMean.cpp line 127 | 21:08 |
-!- travis-ci [~travis-ci@ec2-107-22-23-41.compute-1.amazonaws.com] has joined #shogun | 21:08 | |
travis-ci | [travis-ci] it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/11692011 | 21:08 |
-!- travis-ci [~travis-ci@ec2-107-22-23-41.compute-1.amazonaws.com] has left #shogun [] | 21:08 | |
@HeikoS | votjakovr: ASSERT(m_initialized) | 21:08 |
@HeikoS | so you will have to initialise it | 21:09 |
votjakovr | HeikoS: i probably forgot about init() call | 21:09 |
@HeikoS | votjakovr: maybe a good idea to change this to required and provide an errors message if you have time | 21:09 |
votjakovr | HeikoS: but is it really good idea to call init() manually? | 21:10 |
@sonney2k | votjakovr, yes - e.g. | 21:11 |
@sonney2k | votjakovr, consider you determine variance / mean on some training data | 21:11 |
@sonney2k | and then you only keep a certain number of dims of you data | 21:11 |
@HeikoS | votjakovr: eating, will be back later | 21:11 |
-!- HeikoS [~heiko@218.4.166.178.rev.vodafone.pt] has quit [Quit: Leaving.] | 21:12 | |
@sonney2k | you now want that to work on test data the very same way | 21:12 |
@sonney2k | so you need to have them inited on the training data | 21:12 |
votjakovr | sonney2k: oh yeah, now i see, thanks :) | 21:15 |
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has joined #shogun | 21:21 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 21:22 | |
gsomix | sonney2k, mix of signed/unsigned in one place - pain | 21:26 |
gsomix | sonney2k, https://github.com/shogun-toolbox/shogun/pull/1667 updated | 21:29 |
-!- travis-ci [~travis-ci@ec2-23-22-10-223.compute-1.amazonaws.com] has joined #shogun | 21:30 | |
travis-ci | [travis-ci] it's Soeren Sonnenburg's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/11693769 | 21:30 |
-!- travis-ci [~travis-ci@ec2-23-22-10-223.compute-1.amazonaws.com] has left #shogun [] | 21:30 | |
gsomix | btw I finally changed my gedit+nano environment to the Sublime Text 3. I'm like a hipster now. | 21:31 |
-!- lambday [67157e4c@gateway/web/freenode/ip.103.21.126.76] has joined #shogun | 21:35 | |
lambday | hushell: I am really sorry have been caught up with a few things :( | 21:35 |
-!- zxtx [~zv@173-9-10-22-NewEngland.hfc.comcastbusiness.net] has joined #shogun | 21:51 | |
-!- van51 [~van51@athedsl-409794.home.otenet.gr] has quit [Quit: Leaving.] | 21:55 | |
-!- zxtx [~zv@173-9-10-22-NewEngland.hfc.comcastbusiness.net] has quit [Ping timeout: 248 seconds] | 21:56 | |
-!- travis-ci [~travis-ci@ec2-107-22-23-41.compute-1.amazonaws.com] has joined #shogun | 22:02 | |
travis-ci | [travis-ci] it's Soeren Sonnenburg's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/11693788 | 22:02 |
-!- travis-ci [~travis-ci@ec2-107-22-23-41.compute-1.amazonaws.com] has left #shogun [] | 22:02 | |
-!- lambday [67157e4c@gateway/web/freenode/ip.103.21.126.76] has quit [Ping timeout: 250 seconds] | 22:04 | |
-!- HeikoS [~heiko@218.4.166.178.rev.vodafone.pt] has joined #shogun | 22:14 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 22:14 | |
votjakovr | HeikoS: i've decided to add CWhiteningTransformation to shogun/preprocessor/ | 22:15 |
@HeikoS | votjakovr: awesome! | 22:15 |
@HeikoS | votjakovr: this will be very useful for others, too | 22:15 |
@HeikoS | votjakovr: do we already have the unit variance preprocessor btw? If whitening if too expensive, that one is also useful | 22:16 |
votjakovr | HeikoS: yeah | 22:16 |
votjakovr | HeikoS: PruneVarSubMean | 22:16 |
@HeikoS | votjakovr: I see | 22:17 |
@HeikoS | votjakovr: maybe do a whitening and remove mean combination? its almost always used in conjunction | 22:17 |
votjakovr | HeikoS: oh, yeah, this is not a problem | 22:18 |
@HeikoS | votjakovr: cool! | 22:18 |
@HeikoS | votjakovr: I like that you do that! :) | 22:18 |
* gsomix z-z-z-z | 22:21 | |
@iglesiasg | votjakovr: there is something about whitening in the PCA preprocessor too | 22:21 |
@iglesiasg | votjakovr: maybe you want to check it | 22:22 |
gsomix | nite guys. it seems I need sleep a little | 22:22 |
@HeikoS | gsomix: night! | 22:22 |
@iglesiasg | votjakovr: I took a look yesterday but didn't find it quite similar to the whitening I learnt so... :D | 22:22 |
@HeikoS | iglesiasg: PCA is very similar to whitening | 22:22 |
@iglesiasg | gsomix: good night proto-buffer! :P | 22:22 |
-!- besser82 [~besser82@fedora/besser82] has quit [Quit: freedom, friends, features, first ---> fedoraproject.org] | 22:22 | |
-!- besser82 [~besser82@77-21-40-118-dynip.superkabel.de] has joined #shogun | 22:23 | |
-!- besser82 [~besser82@77-21-40-118-dynip.superkabel.de] has quit [Changing host] | 22:23 | |
-!- besser82 [~besser82@fedora/besser82] has joined #shogun | 22:23 | |
@HeikoS | but a bit more powerful | 22:23 |
votjakovr | iglesiasg: oh, i saw that | 22:23 |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 22:36 | |
shogun-notifier- | shogun: Heiko Strathmann :gp_notebook * 2808d7f / doc/ipython-notebooks/gaussian_process/gaussian_process_classification.ipynb: https://github.com/shogun-toolbox/shogun/commit/2808d7f5b3affdb44d2260e40e8135d617b69f5a | 22:36 |
shogun-notifier- | shogun: some more updates for likelihoods | 22:36 |
* besser82 waves | 22:40 | |
-!- besser82 [~besser82@fedora/besser82] has quit [Quit: freedom, friends, features, first ---> fedoraproject.org] | 22:40 | |
votjakovr | HeikoS: btw i was thinking about mean functions... And i noticed, that it doesn't really metter what type of features we are using for zero mean function, for example | 22:44 |
@HeikoS | votjakovr: yep Zero/Constant mean is independent of the feature type | 22:45 |
@HeikoS | votjakovr: in fact mean functions should be defined on CFeatures objects | 22:45 |
@HeikoS | so you could just change the interface and since we dont have other means, that change would not be a lot of work | 22:45 |
votjakovr | HeikoS: yeah, that what i thought! | 22:45 |
@HeikoS | votjakovr: cool, feel free to change | 22:46 |
@HeikoS | votjakovr: what other things would have to be changed? not many I guess | 22:46 |
@HeikoS | votjakovr: Do you know if GPstuff can do GPs on non-numerical data? | 22:46 |
@HeikoS | votjakovr: I know GPML cannot | 22:46 |
@HeikoS | votjakovr: so would be very very cool to have | 22:46 |
votjakovr | HeikoS: GPstuff also can't | 22:47 |
@HeikoS | votjakovr: killer feature then ;) | 22:47 |
votjakovr | HeikoS: need refactor CMeanFunction a little bit to make it works | 22:48 |
votjakovr | HeikoS: btw how often is GPs used in practice? | 22:50 |
votjakovr | HeikoS: i've found not too much applications | 22:51 |
@HeikoS | votjakovr: ehm ... | 22:51 |
@HeikoS | votjakovr: one of the most fundamental methods in statistics ;) | 22:51 |
@HeikoS | used all over the place in spatio temporal statistics | 22:52 |
@HeikoS | votjakovr: its just not that popular as SVMs, since its a bit harder to understand, but its also more powerful | 22:52 |
@HeikoS | votjakovr: in fact, there is not really a nice impleentation out there | 22:52 |
@HeikoS | votjakovr: but we changed this now partly | 22:52 |
votjakovr | HeikoS: oh :) | 22:54 |
@HeikoS | votjakovr: biologists also use them very often | 22:56 |
@HeikoS | votjakovr: sometimes in different froms, but the core idea is the same | 22:56 |
@HeikoS | votjakovr: we can extend our framework to more cutting edge methods though .... the stuff in there is very fundamental but a bit old-school ;) | 22:56 |
@HeikoS | votjakovr: next thing should some some nice multiclass stuff | 22:56 |
@HeikoS | votjakovr: and then of course sparsity extensions? aka relevance vector machine | 22:57 |
@HeikoS | they scale linearly in the nubmer of training points | 22:57 |
votjakovr | HeikoS: cool :) | 22:59 |
-!- travis-ci [~travis-ci@ec2-23-22-10-223.compute-1.amazonaws.com] has joined #shogun | 23:11 | |
travis-ci | [travis-ci] it's Soeren Sonnenburg's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/11694148 | 23:11 |
-!- travis-ci [~travis-ci@ec2-23-22-10-223.compute-1.amazonaws.com] has left #shogun [] | 23:11 | |
@iglesiasg | mmm I guess this week there should be weekly report mail as well, right? | 23:15 |
votjakovr | iglesiasg: hmm it seems nobody have sent a report | 23:17 |
@HeikoS | no need for a report, send you notebooks ;) | 23:18 |
votjakovr | HeikoS: notebook is still in process | 23:19 |
@iglesiasg | HeikoS: all right! | 23:19 |
@HeikoS | votjakovr: looking forward to see | 23:19 |
votjakovr | :) | 23:19 |
@iglesiasg | I sent a new update yesterday, but I think I am going to do a small thing now before going to bed | 23:20 |
@HeikoS | votjakovr: I have written some more things, pushing now | 23:20 |
@iglesiasg | HeikoS: aah true you already saw and commented what I pushed yesterday, thanks btw :) | 23:20 |
shogun-notifier- | shogun: Heiko Strathmann :gp_notebook * 6fa30fc / doc/ipython-notebooks/gaussian_process/gaussian_process_classification.ipynb: https://github.com/shogun-toolbox/shogun/commit/6fa30fc22a6f8936b550e34c22211eee0d7a1cd4 | 23:20 |
shogun-notifier- | shogun: another paragraph on inference for binary classification | 23:20 |
@HeikoS | iglesiasg: welcome :) | 23:21 |
@HeikoS | iglesiasg: feedback is very important for the notebooks, even more than for code | 23:21 |
@HeikoS | iglesiasg: if you can give some more feedback to others, in particular those who haven't received some yet | 23:21 |
votjakovr | HeikoS: oh thanks, rebasing... | 23:21 |
@iglesiasg | HeikoS: all right | 23:22 |
@iglesiasg | HeikoS: I started reading lambday's notebook yesterday | 23:22 |
@HeikoS | iglesiasg: cool, also have a look at van51s | 23:23 |
@iglesiasg | HeikoS: but I have to admit I quit after some minutes and left it for another moment | 23:23 |
@HeikoS | iglesiasg: I gave lambday a *lot* of feedback already, also the stats prof here and the other mentor did | 23:23 |
@iglesiasg | HeikoS: yeah, already checked van's and told him one minor thing | 23:23 |
@HeikoS | iglesiasg: try to focus on having things look very cool | 23:24 |
@iglesiasg | brb | 23:41 |
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 23:41 | |
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has joined #shogun | 23:44 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 23:44 | |
--- Log closed Tue Sep 24 00:00:12 2013 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!