--- Log opened Mon Feb 06 00:00:13 2017 | ||
@sukey | Pull Request #3608 "Clean up KNN" synchronized by MikeLing - https://github.com/shogun-toolbox/shogun/pull/3608 | 01:31 |
---|---|---|
@sukey | Pull Request #3584 "LinalgRefactor - Apply" synchronized by OXPHOS - https://github.com/shogun-toolbox/shogun/pull/3584 | 07:40 |
-!- mikeling [uid89706@gateway/web/irccloud.com/x-kznnvwkxtejgctvi] has joined #shogun | 08:28 | |
-!- praisethemoon [~praisethe@197.3.11.178] has joined #shogun | 10:37 | |
-!- praisethemoon [~praisethe@197.3.11.178] has quit [Ping timeout: 248 seconds] | 11:12 | |
-!- praisethemoon [~praisethe@197.3.11.178] has joined #shogun | 12:01 | |
-!- praisethemoon [~praisethe@197.3.11.178] has quit [Changing host] | 12:02 | |
-!- praisethemoon [~praisethe@unaffiliated/praisethemoon] has joined #shogun | 12:02 | |
-!- HeikoS [~heiko@untrust-out.swc.ucl.ac.uk] has joined #shogun | 12:05 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 12:05 | |
-!- ShellcatZero [~ShellcatZ@cpe-76-93-136-67.san.res.rr.com] has quit [Ping timeout: 245 seconds] | 12:07 | |
@wiking | lisitsyn, pppppppppuuuung | 12:30 |
lisitsyn | wiking: peng | 12:35 |
@wiking | ok | 12:45 |
@wiking | org.shogun-toolbox.* or ml.shogun.* | 12:45 |
@wiking | as java pkgs? | 12:45 |
-!- praisethemoon [~praisethe@unaffiliated/praisethemoon] has quit [Ping timeout: 248 seconds] | 12:56 | |
CaBa | wiking: heya | 12:57 |
CaBa | oh | 12:57 |
CaBa | HeikoS: hey, you happen to be around? | 12:58 |
@HeikoS | CaBa: hi yes | 12:58 |
CaBa | HeikoS: remember when we were talking the other day about kernel functions that depend not only on something that cannot be plugged into a CFeature class, but for example index structures and things to speed up the computation of K(x_1, x_2)? And you said one solution would be to precalc the kernel matrix and use CCustomKernel but that that would not support e.g. normalizers? | 13:00 |
@HeikoS | yes | 13:00 |
@HeikoS | i remember | 13:00 |
@wiking | CaBa, yes | 13:00 |
@wiking | hi | 13:01 |
@HeikoS | wiking: my plan for today: write GSoc projects | 13:01 |
@HeikoS | wiking: you got some bullet points on that? | 13:01 |
@HeikoS | same projects as last year more or less? ;) | 13:01 |
CaBa | HeikoS: i know this is dirty, but would this work: use CDenseFeatures<size_t>, 1-dimensional, the size_t value actually being an index for a feature vector... then derive from CKernel, override ::compute(), which then performs lookups according to the size_t feature in the "real" kernel matrix and returns the value. that would be a fully functional kernel, wouldnt it? one that supports subsetting and therefor | 13:04 |
CaBa | e x-val, normalization during x-val, MKL, etc? no restrictions? | 13:04 |
@HeikoS | yes that would work | 13:06 |
@HeikoS | but it is a hack | 13:06 |
@HeikoS | as your kernel would only work with this very specialized features | 13:06 |
@HeikoS | generally, if you implement subsetting on kernel and features, then you can do anything you like and it will work | 13:07 |
@HeikoS | "anything you like" as in override methods in a weird way | 13:07 |
CaBa | HeikoS: well the features wouldn't be visible to the user - just like in CCustomKernel. The kernel would just take a precomputed matrix in it's constructor. | 13:07 |
@HeikoS | sure | 13:07 |
CaBa | HeikoS: it would be kind of a hacky replacement for CCustomKernel that supports normalization | 13:08 |
@HeikoS | what I mean is that you have the inheritance diagram, and there two leafs (i.e. specialized classes) depend on one another | 13:08 |
@HeikoS | which is against OOP paradigma | 13:08 |
@HeikoS | but yeah | 13:08 |
@HeikoS | I see your need for hacks ;) | 13:08 |
@HeikoS | would be a cool thing to do as part of a GSoC project on x-validation and modelselection | 13:09 |
CaBa | my need for hacks? ^^ | 13:09 |
@HeikoS | you seem to have this complicated kernel | 13:10 |
@HeikoS | btw did you ever think about just not normalising the x-validation kernel matrices? | 13:10 |
@HeikoS | i.e. just normalise your precomputed kernel matrix? | 13:10 |
CaBa | sure, but it's a different thing | 13:10 |
@HeikoS | or is it really crucial? | 13:11 |
CaBa | i don't know if it's relevant | 13:11 |
@HeikoS | I could also btw imagine some on-the-fly normalisation that works with custom kernels | 13:11 |
@wiking | HeikoS, ah man | 13:11 |
@wiking | i have to remember all those good ideas | 13:11 |
@wiking | we have in the parking lot | 13:11 |
@wiking | :D | 13:11 |
@HeikoS | wiking: yes that is what I am asking :) | 13:11 |
@HeikoS | wiking: I have a few things | 13:11 |
@HeikoS | but before I tell them, I wanted to ask to get more coverage | 13:11 |
@wiking | HeikoS, btw we need to rename our java package... it's either org.shogun-toolbox.* or ml.shogun.* | 13:12 |
@wiking | as we do not own org.shogun | 13:12 |
@wiking | i mean shogun.org | 13:12 |
@HeikoS | uh, nice | 13:12 |
CaBa | btw - do you know whether per-kernel sqrt-diag normalization is sufficient in MKL to get interpretable subkernel weights? or if other means of normalization are necessary? | 13:12 |
@wiking | HeikoS, so we need to make a decision here... | 13:12 |
@HeikoS | CaBa: for xvalidation, you dont need interpretable weights or do you? | 13:12 |
@HeikoS | wiking: ml.shogun | 13:13 |
@wiking | ok | 13:13 |
@HeikoS | CaBa: or just with the final model | 13:13 |
@wiking | i'll register | 13:13 |
@wiking | lisitsyn, ^ | 13:13 |
@HeikoS | wiking: cool! | 13:13 |
@HeikoS | CaBa: because if you just tune parameters using xvalidation, then maybe it is not that important | 13:13 |
CaBa | HeikoS: that's what i'm wondering - will i just get different betas if i normalize afterwards but the subkernel parameters remain optimal? | 13:14 |
@HeikoS | CaBa: probably it is worth a try | 13:14 |
@HeikoS | but I dont know much about MKL | 13:15 |
@HeikoS | and sonney2k who is the only one who knows, is busy with life | 13:15 |
CaBa | HeikoS: life is a cool thing to be busy with ;) | 13:18 |
@HeikoS | CaBa: so just to get you right | 13:18 |
@HeikoS | CaBa: you want | 13:18 |
@HeikoS | to use | 13:18 |
@HeikoS | custom kernel + mkl + cross-validation + kernel normaliser | 13:19 |
@HeikoS | maybe we can just make this an entrance task | 13:19 |
@HeikoS | we got some students incoming for GSoC | 13:19 |
@HeikoS | and this might be a nice non-trivial thing to work out | 13:19 |
CaBa | that sounds about right | 13:19 |
@HeikoS | so what we need here is that the kernel normalisation works with the subsetting and the mkl, all on the fly | 13:20 |
CaBa | well... they hacky replacement for CCustomKernel I had in mind is a 30 min job I guess. But I'm sure you had something clean in mind :D | 13:20 |
@HeikoS | yes sure, I mean for your purposes, the 30min hack might be ok | 13:20 |
@HeikoS | but It is good for us to get such things done properly, in case somebody else also wants to use this kind of thing | 13:21 |
CaBa | The only ting CCustomKernel lacks in that respect might be the normalization. Last time we spoke you said x-val and mkl might just work. | 13:21 |
CaBa | But I didn't check that I must admit | 13:21 |
lisitsyn | wiking: ml.shogun if we don't lost it again | 13:24 |
lisitsyn | :) | 13:24 |
@wiking | lisitsyn, i've bought it for 5 years :) | 13:25 |
lisitsyn | org.shogun-toolbox is impossible | 13:25 |
lisitsyn | would be | 13:25 |
lisitsyn | org.shogun_toolbox | 13:25 |
lisitsyn | superfuckingugly | 13:25 |
@wiking | yes | 13:26 |
CaBa | SG_REF checks for nullptr, right? | 13:31 |
@HeikoS | yes | 13:32 |
@HeikoS | but best read the def | 13:32 |
CaBa | what's that deprecated info in the wiki? is there a replacement? | 13:32 |
@HeikoS | ? | 13:33 |
@HeikoS | ah | 13:33 |
@HeikoS | no :) | 13:33 |
CaBa | ^^ | 13:33 |
@sukey | Pull Request #3608 "Clean up KNN" synchronized by MikeLing - https://github.com/shogun-toolbox/shogun/pull/3608 | 14:10 |
mikeling | hi, i got a bunch of undocument warning and failed to build python modular https://pastebin.mozilla.org/8974089 | 14:41 |
-!- praisethemoon [~praisethe@197.3.11.178] has joined #shogun | 14:56 | |
mikeling | wiking: ping | 15:20 |
@wiking | pong | 15:22 |
mikeling | wiking: after I pull docker image, how could I use python interface? | 15:22 |
mikeling | on the image i mean | 15:23 |
@wiking | - docker run -t -d -P -e "JAVA_HOME=/usr/lib/jvm/java-8-oracle" -e "CC=$CC" -e "CXX=$CXX" --name devenv -v $PWD:/opt/shogun shogun/shogun-dev /bin/sh -c "mkdir /opt/shogun/build;bash" | 15:23 |
@wiking | - docker exec -t devenv /bin/sh -c "cd /opt/shogun/build; cmake -DCMAKE_INSTALL_PREFIX=$HOME/shogun-build -DENABLE_TESTING=ON $CMAKE_OPTIONS .." | 15:23 |
@wiking | where CMAKE_OPTIONS="-DPythonModular=ON -DTRAVIS_DISABLE_UNIT_TESTS=ON -DTRAVIS_DISABLE_LIBSHOGUN_TESTS=ON" | 15:23 |
mikeling | and pip install shogun-ml doesn't work, it looks like there is no package named shogun-ml? | 15:24 |
@wiking | pip install? | 15:26 |
@wiking | why do you want pip install when you try a local development | 15:26 |
mikeling | i just see that on document and want to have a try | 15:26 |
@wiking | ah forget it | 15:27 |
@wiking | it needs further fixing | 15:27 |
@wiking | hopefully 5.1.0 will have it | 15:27 |
mikeling | ok ;) | 15:27 |
mikeling | wiking: It says "CMake Error: The source directory "/opt/shogun" does not appear to contain CMakeLists.txt. Specify --help for usage, or press the help button on the CMake GUI." | 15:50 |
@wiking | ok | 15:50 |
@wiking | look i'll write this the one time | 15:50 |
@wiking | docker run -t -i -v <the path to your checked out source>:/opt/shogun shogun/shogun-dev /bin/bash | 15:51 |
@wiking | (now you are in docker) | 15:51 |
@wiking | cd /opt/shogun | 15:51 |
@wiking | mkdir build | 15:51 |
@wiking | cd build | 15:51 |
@wiking | cmake .... (all the stuff above) | 15:51 |
@wiking | make | 15:51 |
@HeikoS | mikeling: is there anything that is not documented? | 15:54 |
mikeling | HeikoS: yep, part of warning is in https://pastebin.mozilla.org/8974116 | 15:55 |
@HeikoS | I mean as in documented installation process on our website | 15:56 |
@HeikoS | these are just warnings | 15:56 |
mikeling | HeikoS: oh, the pip install shogun-ml doesn't work | 15:56 |
mikeling | it seems like shogun package is missing | 15:56 |
@HeikoS | mikeling: I think I dont understands what you are actually doing there. these were just 3 very unrelated things (docker, pip install, cmake) | 15:59 |
mikeling | HeikoS: because I would like to install the python interface of shogun, after I failed to build it from source code locally, I switch to the pip install but it looks like missing shogun-ml package for it | 16:01 |
mikeling | and now, I want to use docker to solver my problem | 16:01 |
mikeling | BTW, the error message is /bin/sh: line 1: 13156 Segmentation fault: 11 /usr/local/bin/doxygen modshogun.doxy when I build python interface locally | 16:04 |
-!- HeikoS [~heiko@untrust-out.swc.ucl.ac.uk] has quit [Ping timeout: 240 seconds] | 16:05 | |
mikeling | And here is the Cmake I used " cmake -DPYTHON_INCLUDE_DIR=/usr/local/Cellar/python/2.7.13/Frameworks/Python.framework/Headers -DPYTHON_LIBRARY=/usr/local/Cellar/python/2.7.13/Frameworks/Python.framework/Versions/2.7/lib/libpython2.7.dylib -DPythonModular=ON ../" | 16:05 |
@wiking | mikeling, ? | 16:08 |
@wiking | wtf | 16:08 |
@wiking | i told you how you can do it with the dokcer | 16:08 |
@wiking | but you can do the same way on your osx | 16:08 |
@wiking | cmake -DPythonModular=ON .. | 16:09 |
@wiking | make | 16:09 |
@wiking | either way works | 16:09 |
mikeling | wiking: yeah, I can do the same thing on osx, but it doesn't work for me, at least it failed on locally without docker | 16:12 |
@wiking | what's the failure? | 16:12 |
@wiking | btw without your patches | 16:12 |
@wiking | or with your patches? | 16:12 |
mikeling | without my pr | 16:13 |
mikeling | s /bin/sh: line 1: 13156 Segmentation fault: 11 /usr/local/bin/doxygen modshogun.doxy | 16:13 |
mikeling | without the "s", sorry | 16:13 |
@wiking | brew info doxygen | 16:13 |
mikeling | doxygen: stable 1.8.13 (bottled), HEAD | 16:14 |
@wiking | git git rev-parse HEAD | 16:16 |
@wiking | ? | 16:16 |
@wiking | but in worst case you can disable doxygen and do it without that | 16:17 |
mikeling | you mean git rev-parse HEAD? | 16:17 |
@wiking | yes | 16:18 |
mikeling | it's 9673755bdb345b3c10fe447425394e4813f06dd8 | 16:18 |
@wiking | mmm i've just ran on my osx | 16:18 |
@wiking | and i dont get segfault | 16:18 |
mikeling | and here is my cmake output https://pastebin.mozilla.org/8974123 | 16:18 |
@wiking | -- Found NumPy: version "1.12.0" /usr/local/Cellar/matplotlib/2.0.0/libexec/lib/python2.7/site-packages/numpy/core/include? | 16:18 |
@wiking | this doesn't look too good | 16:19 |
mikeling | hmmm, so you think it may case a segfault? | 16:20 |
@wiking | noup | 16:20 |
@wiking | bu ti dont understand | 16:20 |
@wiking | why the numpy is part of matplotlib | 16:20 |
@wiking | pip list|grep numpy | 16:20 |
@wiking | ? | 16:20 |
mikeling | it return numpy (1.12.0) | 16:20 |
@wiking | pip show numpy | 16:21 |
@wiking | ? | 16:21 |
mikeling | do you think docker can help me solve that segfault? | 16:22 |
@wiking | with docker you would get exactly the same thing | 16:22 |
@wiking | what happens on travis | 16:22 |
@wiking | as it's the same env | 16:22 |
mikeling | https://www.irccloud.com/pastebin/keyKp2ID/ | 16:22 |
@wiking | yeah i have no idea why it is landing in Location: /usr/local/Cellar/matplotlib/2.0.0/libexec/lib/python2.7/site-packages | 16:23 |
@wiking | but ok... | 16:23 |
mikeling | so I just working around this segfault for whole day and no idea why it happened and how to fix it | 16:24 |
@wiking | brew uninstall doxygen | 16:26 |
@wiking | and try it w/o that | 16:26 |
mikeling | ok, let me have a try | 16:26 |
mikeling | wiking: Oh my mine, it works....... | 16:36 |
mikeling | after remove doxygen | 16:36 |
mikeling | thank you very much | 16:39 |
-!- suhas2go [uid201652@gateway/web/irccloud.com/x-ukivykjvzmoynhgd] has joined #shogun | 17:08 | |
@sukey | New branch feature/OutOfSource created on shogun-toolbox/shogun | 17:11 |
@sukey | New Commit "Remove all none out of source patterns in build" to shogun-toolbox/shogun by vigsterkr: https://github.com/shogun-toolbox/shogun/commit/922e3ab6494917b5767a1d6bb4e5740854c4216e | 17:12 |
-!- ShellcatZero [~ShellcatZ@cpe-76-93-136-67.san.res.rr.com] has joined #shogun | 18:10 | |
-!- HeikoS [~heiko@untrust-out.swc.ucl.ac.uk] has joined #shogun | 18:15 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 18:15 | |
-!- travis-ci [~travis-ci@ec2-54-162-62-3.compute-1.amazonaws.com] has joined #shogun | 18:15 | |
travis-ci | it's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/198887697 | 18:15 |
-!- travis-ci [~travis-ci@ec2-54-162-62-3.compute-1.amazonaws.com] has left #shogun [] | 18:15 | |
-!- praisethemoon [~praisethe@197.3.11.178] has quit [Ping timeout: 248 seconds] | 18:33 | |
-!- mikeling [uid89706@gateway/web/irccloud.com/x-kznnvwkxtejgctvi] has quit [Quit: Connection closed for inactivity] | 19:09 | |
-!- HeikoS [~heiko@untrust-out.swc.ucl.ac.uk] has quit [Ping timeout: 252 seconds] | 21:28 | |
--- Log closed Tue Feb 07 00:00:14 2017 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!