| --- Log opened Wed Mar 30 00:00:36 2011 | ||
| -!- aifargonos [~aifargono@46.18.27.35] has quit [Ping timeout: 260 seconds] | 00:30 | |
| -!- jabbok [8d550074@gateway/web/freenode/ip.141.85.0.116] has quit [Quit: Page closed] | 00:52 | |
| -!- Occam_ [8182c749@gateway/web/freenode/ip.129.130.199.73] has joined #shogun | 01:00 | |
| -!- Occam_ [8182c749@gateway/web/freenode/ip.129.130.199.73] has quit [Client Quit] | 01:00 | |
| -!- shelhamer [~shelhamer@AMontsouris-152-1-50-9.w83-202.abo.wanadoo.fr] has quit [Quit: Get MacIrssi - http://www.sysctl.co.uk/projects/macirssi/] | 01:23 | |
| -!- epps [~epps@unaffiliated/epps] has quit [Ping timeout: 246 seconds] | 02:50 | |
| -!- alesis-novik [~alesis@188.74.87.84] has joined #shogun | 05:11 | |
| -!- Ryaether [~Ryaether@50-80-170-245.client.mchsi.com] has joined #shogun | 05:15 | |
| -!- Ryaether [~Ryaether@50-80-170-245.client.mchsi.com] has quit [Client Quit] | 05:16 | |
| -!- Ryaether [~Ryaether@50-80-170-245.client.mchsi.com] has joined #shogun | 05:16 | |
| -!- aifargonos [~aifargono@46.18.27.35] has joined #shogun | 07:51 | |
| -!- Tanmoy [75d35896@gateway/web/freenode/ip.117.211.88.150] has joined #shogun | 07:59 | |
| -!- siddharth__ [~siddharth@117.211.88.150] has joined #shogun | 08:18 | |
| -!- siddharth_ [~siddharth@117.211.88.150] has quit [Ping timeout: 252 seconds] | 08:20 | |
| -!- aifargonos [~aifargono@46.18.27.35] has quit [Ping timeout: 240 seconds] | 08:34 | |
| -!- meanerelk [~meanerelk@cpe-75-187-51-178.columbus.res.rr.com] has quit [Ping timeout: 240 seconds] | 08:39 | |
| -!- aifargonos [~aifargono@193.206.186.107] has joined #shogun | 08:45 | |
| -!- aifargonos [~aifargono@193.206.186.107] has left #shogun [] | 08:52 | |
| -!- siddharth_ [~siddharth@117.211.88.150] has joined #shogun | 09:00 | |
| -!- aifargonos [~aifargono@193.206.186.107] has joined #shogun | 09:01 | |
| -!- siddharth__ [~siddharth@117.211.88.150] has quit [Ping timeout: 264 seconds] | 09:03 | |
| -!- siddharth__ [~siddharth@117.211.88.150] has joined #shogun | 09:07 | |
| -!- siddharth_ [~siddharth@117.211.88.150] has quit [Ping timeout: 246 seconds] | 09:10 | |
| -!- Ryaether [~Ryaether@50-80-170-245.client.mchsi.com] has left #shogun [] | 09:11 | |
| -!- jabbok [56220ef5@gateway/web/freenode/ip.86.34.14.245] has joined #shogun | 09:31 | |
| -!- aifargonos [~aifargono@193.206.186.107] has quit [Quit: aifargonos] | 09:45 | |
| -!- aifargonos [~aifargono@193.206.186.107] has joined #shogun | 09:46 | |
| -!- blackburn [~qdrgsm@188.168.4.226] has joined #shogun | 10:01 | |
| -!- blackburn [~qdrgsm@188.168.4.226] has quit [Ping timeout: 252 seconds] | 10:11 | |
| -!- blackburn [~qdrgsm@188.168.4.148] has joined #shogun | 10:11 | |
| -!- aifargonos [~aifargono@193.206.186.107] has quit [Ping timeout: 250 seconds] | 10:12 | |
| -!- siddharth__ [~siddharth@117.211.88.150] has quit [Ping timeout: 240 seconds] | 10:20 | |
| -!- aifargonos [~aifargono@46.18.27.4] has joined #shogun | 10:37 | |
| -!- blackburn [~qdrgsm@188.168.4.148] has quit [Quit: Leaving.] | 10:45 | |
| -!- epps [~epps@unaffiliated/epps] has joined #shogun | 12:14 | |
| -!- yin [~chatzilla@207.46.92.18] has quit [Read error: Connection reset by peer] | 12:27 | |
| -!- aifargonos [~aifargono@46.18.27.4] has quit [Ping timeout: 276 seconds] | 12:28 | |
| -!- epps [~epps@unaffiliated/epps] has quit [Quit: Leaving] | 13:07 | |
| -!- siddharth__ [~siddharth@117.211.88.150] has joined #shogun | 13:10 | |
| -!- aifargonos [~aifargono@46.18.27.35] has joined #shogun | 13:17 | |
| -!- CIA-30 [~CIA@208.69.182.149] has joined #shogun | 13:46 | |
| sonney2k | CIA finally arrived :D | 13:46 |
|---|---|---|
| sonney2k | all hail our chatty bot :) | 13:46 |
| -!- shelhamer [~shelhamer@gut75-5-82-247-9-141.fbx.proxad.net] has joined #shogun | 13:47 | |
| CIA-30 | shogun: Soeren Sonnenburg master * rf9421d3 / .gitignore : update gitignore to exclude looking for ui symlink - http://bit.ly/eweRGm | 13:48 |
| CIA-30 | shogun: Soeren Sonnenburg master * re4450d2 / src/configure : | 13:48 |
| CIA-30 | shogun: add flags to support lua,ruby and java via | 13:48 |
| CIA-30 | shogun: ./configure --interfaces=libshogun,ruby_modular,lua_modular,java_modular | 13:48 |
| CIA-30 | shogun: basic autodetection should work for java and ruby paths. Autodetection | 13:48 |
| CIA-30 | shogun: for lua should be perfect. - http://bit.ly/fMbsF8 | 13:48 |
| CIA-30 | shogun: Soeren Sonnenburg master * r1ed9e90 / (12 files): update modular interface files to discard SWIG_fail for java and doxygen includes for ruby - http://bit.ly/hyBoNs | 13:48 |
| -!- shelhamer [~shelhamer@gut75-5-82-247-9-141.fbx.proxad.net] has quit [Client Quit] | 13:49 | |
| -!- shelhamer [~shelhamer@gut75-5-82-247-9-141.fbx.proxad.net] has joined #shogun | 13:49 | |
| aifargonos | nice | 13:49 |
| sonney2k | aifargonos, github just rocks... | 13:50 |
| aifargonos | I just wonder why I did not switch to git muuuuuch earlier .. | 13:51 |
| sonney2k | aifargonos, well I do so too... it is so nice to just register with github and then 'fork a project' and auto track it... | 13:55 |
| shelhamer | git+github is about as convenient as distributed coding can get. I'm happy to see the code move over from svn | 13:59 |
| shelhamer | sonney2k, I am a potential GSoC student but I have a concern about another commitment I have in the summer, and wanted to ask about it now (assuming this is the appropriate forum for it–if not, I can email you) | 14:00 |
| sonney2k | shoot | 14:05 |
| shelhamer | I am an AI-concentration CS undergrad in the US (I'll be in my senior year in the fall) and this summer I will be working for the computer vision lab, as I have done in past summers | 14:06 |
| shelhamer | this means that I would have about 20 hours a week for shogun, were I to apply as GSoC | 14:07 |
| shelhamer | do you believe that would be enough, or would it be insufficient? | 14:07 |
| shelhamer | I understand completely if that does not meet the time commitment you expect, but wanted to ask as I have some machine learning background and solid mathematical foundations from vision | 14:07 |
| sonney2k | shelhamer, double commitments are tough. google suggests that this is a no-go http://google-opensource.blogspot.com/2011/03/dos-and-donts-of-google-summer-of-code.html | 14:09 |
| shelhamer | yes, I had read that, but feel that I may be on the borderline as last summer I worked for the lab and my own programming business through the summer with success; I have in fact paid for my schooling by programming on the side all along | 14:10 |
| shelhamer | but at the same time, I know 20 != 40 | 14:10 |
| sonney2k | shelhamer, I have put similarly high workload on myself (it usually does not stay at 20 for each of your jobs) and it can ruin the fun | 14:10 |
| shelhamer | this is true, I've had similar experiences myself, with one job at 40 and another at 10-20, and don't want to repeat it, but found the prospect of GSoC exciting | 14:11 |
| shelhamer | with your saying that though, I suppose I should wait until next year | 14:11 |
| sonney2k | As much as I am positive about it (due to your background) I would really want to warn you... | 14:12 |
| shelhamer | all the same, I'll plan on getting to know some of the shogun code this summer, just unofficially, and maybe I'll be able to help out in some way | 14:12 |
| shelhamer | I appreciate your warning, and thank you for your advice! as you say, I don't want to have it turn out miserably from over-committing | 14:12 |
| sonney2k | that would definitely turn you into a strong candidate next year ... | 14:13 |
| shelhamer | well, that's the way forward then-thank you for helping me decide. I'll be graduating next spring, so should have more freedom that summer to participate in something like this | 14:14 |
| shelhamer | thanks again | 14:14 |
| sonney2k | shelhamer, you are welcome and thanks for your interest! all feedback is welcome. | 14:15 |
| -!- heiko [~strathman@neckar.uni-duisburg.de] has joined #shogun | 14:19 | |
| -!- bla_ [c07c1afa@gateway/web/freenode/ip.192.124.26.250] has joined #shogun | 14:53 | |
| -!- bla_ [c07c1afa@gateway/web/freenode/ip.192.124.26.250] has quit [Client Quit] | 14:54 | |
| @knrrrd | hi dudes | 15:09 |
| sonney2k | ohh the return of the ueberdude | 15:10 |
| @knrrrd | Hehe. I am Bonn HBF waiting for next connection | 15:11 |
| sonney2k | hail to the german trains, hail the deutsche bahn :-( | 15:11 |
| -!- mode/#shogun [+o sonney2k] by ChanServ | 15:12 | |
| @knrrrd | mobile IRC is really entertaining | 15:13 |
| @sonney2k | so that is from your iphone? | 15:14 |
| @knrrrd | Yep | 15:14 |
| @sonney2k | I guess bci controlled. | 15:14 |
| @knrrrd | Almost | 15:15 |
| @sonney2k | that cap will attract extra attention | 15:15 |
| @knrrrd | Next stop | 15:15 |
| @knrrrd | Cologne | 15:15 |
| aifargonos | sonney2k: thanks for email reply ... | 15:20 |
| @sonney2k | aifargonos, you are welcome... | 15:21 |
| aifargonos | I guess I'll apply for this hdf5 idea ... | 15:21 |
| aifargonos | hdf5 looks like coor format ... | 15:21 |
| aifargonos | what functionality is already implemented ?? | 15:21 |
| @sonney2k | there is lots of things to do... | 15:22 |
| @sonney2k | have a look at CHDF5File.{cpp,h} | 15:22 |
| aifargonos | (I've just opened it :-) ) | 15:22 |
| @sonney2k | we can read/write vectors/matrices of any type, strings is missing, symlinks are missing and it would be great to be able to read newer matlab files (they are hdf5 based) | 15:23 |
| @sonney2k | and also pytables based files | 15:23 |
| @sonney2k | I guess also some custom functions like create a group / change into that group / list the file content... | 15:24 |
| aifargonos | where can I find doc for this matlab and py stuff ?? | 15:24 |
| aifargonos | or I'll just google it ... | 15:24 |
| @sonney2k | aifargonos, pytables you can probably find, matlab - I guess we will have to supply some samples | 15:25 |
| @sonney2k | (it is not documented) | 15:25 |
| aifargonos | :-( | 15:25 |
| @sonney2k | closed source at its best | 15:25 |
| aifargonos | .. so you want just general functions for file manipulation ?? | 15:26 |
| aifargonos | I thought it was meant to be a form of serialization ... | 15:26 |
| @sonney2k | aifargonos, serialization is already there | 15:26 |
| @sonney2k | even xml, ascii, hdf5 and json based | 15:26 |
| aifargonos | ok ..: I'll have look at it ... | 15:27 |
| @sonney2k | I'd like to simply be able to load lots of the hdf5 based formats | 15:27 |
| @sonney2k | I mean hdf5 is just a container format... | 15:27 |
| @sonney2k | btw, mldata.org format support too... | 15:27 |
| aifargonos | heh .. ok, so you want handles for a manipulation of different data formats .. yes ?? | 15:29 |
| aifargonos | or even extending serialisation ?? | 15:29 |
| aifargonos | or I don't need to care about serialization ?? | 15:29 |
| @sonney2k | aifargonos, you don't need to care about serialization ... I care about being able to read these different hdf5 based data formats and to be able to write all of shogun's features in hdf5 | 15:32 |
| -!- seviyor [c1e20418@gateway/web/freenode/ip.193.226.4.24] has quit [Quit: Page closed] | 15:35 | |
| aifargonos | ok .: so then, what about some common interface for manipulating whatever format will be supported ... some kind of abstraction layer ..: you use same calls for work with different formats, where concrete format depends on some setting / your choice .. ?? | 15:36 |
| @sonney2k | aifargonos, Heh - that is what we have in CFile.h :) | 15:39 |
| @sonney2k | but yes there is a lot of work that can be done here... better .csv support, etc... | 15:40 |
| aifargonos | cool .. ok so I'll go through the code tonight to get better idea about it ... | 15:40 |
| -!- blackburn [~qdrgsm@188.168.4.243] has joined #shogun | 16:09 | |
| -!- shelhamer [~shelhamer@gut75-5-82-247-9-141.fbx.proxad.net] has quit [Quit: Computer has gone to sleep] | 16:23 | |
| -!- aifargonos [~aifargono@46.18.27.35] has quit [Ping timeout: 260 seconds] | 16:55 | |
| -!- seviyor [c1e20418@gateway/web/freenode/ip.193.226.4.24] has joined #shogun | 16:56 | |
| -!- blackburn [~qdrgsm@188.168.4.243] has quit [Ping timeout: 276 seconds] | 16:59 | |
| -!- blackburn [~qdrgsm@188.168.2.29] has joined #shogun | 17:00 | |
| -!- sploving [~root@124.16.139.196] has quit [Ping timeout: 240 seconds] | 17:01 | |
| -!- sploving [~root@124.16.139.196] has joined #shogun | 17:01 | |
| -!- aifargonos [~aifargono@193.206.186.107] has joined #shogun | 17:24 | |
| jabbok | @sonney2k, I have a question regarding COFFIN, in the 2010 ICML paper you say that in the first example you used features with 141 attributes and in the second (gender estimation) you use 2827 attributes, all equal to one in a 723,712 vector, the rest being zero | 17:44 |
| jabbok | @sonney2k, have you tried COFFIN on other features, with higher dimensionalities, such as Histogram of Gradients or something like that with ~2000 attributes, non-spare, all real ? | 17:46 |
| @sonney2k | jabbok, no I didn't | 17:49 |
| @sonney2k | I don't think that this would be particularly fast though | 17:49 |
| jabbok | the DOT and ADD operations were overwritten in liblinear original code? | 17:51 |
| jabbok | or the principles from liblinear were reimplemented in COFFIN? | 17:52 |
| -!- dadkins [~kvirc@cpe-174-102-36-192.woh.res.rr.com] has joined #shogun | 17:53 | |
| @sonney2k | jabbok overwritten in liblinear, SGD, ocas ... | 17:54 |
| jabbok | I've setup the shogun environment on a ubuntu 10.10 from the git repository; now cmdline, octave and matlab both static and modular work | 17:56 |
| jabbok | but I couldn't find any liblinear after a "find . -name liblinear" command, cout you please give me a hint where they might be, in case they are in the git repository ? | 17:57 |
| @sonney2k | find . -iname '*liblinear*' | 17:57 |
| @sonney2k | IIRC LibLinear.h | 17:58 |
| jabbok | sorry, my bad, online doxygen was not case sensitive also | 18:00 |
| dadkins | I have some short gsoc-related questions: any takers? | 18:03 |
| @sonney2k | shoot | 18:03 |
| dadkins | I was able to get the latest trunk and got everything building easily enough. For my project I plan to integrate C5.0 and kPCA: does this seem like a good course of action for a proposal? | 18:05 |
| dadkins | let me be more clear about "integrate C5.0 and kPCA"... | 18:06 |
| @sonney2k | dadkins, there are quite a number of takers for C5.0 and kPCA is basically implemented but only needs some polishing | 18:06 |
| @sonney2k | so it would be better to think of an alternative | 18:07 |
| dadkins | any suggestions? | 18:07 |
| @sonney2k | I would recommend to integrate some algorithm (for which an implementation exists) and to implement something that is not there. but what exactly depends on what you would be interested in. | 18:08 |
| @sonney2k | something that you might want to use yourself for example... | 18:08 |
| blackburn | but not of dimensionality reduction :) | 18:09 |
| blackburn | *joke* | 18:09 |
| -!- aifargonos [~aifargono@193.206.186.107] has quit [Ping timeout: 250 seconds] | 18:09 | |
| -!- heiko [~strathman@neckar.uni-duisburg.de] has left #shogun [] | 18:09 | |
| jabbok | @sonney2k, all that's in shogun for liblinear are contended in LibLinear.h and LibLinear.cpp, right ? | 18:10 |
| jabbok | and COFFIN is fully integrated in shogun without any other standalone sources? | 18:11 |
| @sonney2k | jabbok, I think there is also some other files like Tron.{cpp,h} but for e.g. the linear L1-loss SVM yes | 18:11 |
| @sonney2k | jabbok, well the dotfeatures provide the needed functions | 18:11 |
| @sonney2k | so I only had to replace things like: add alpha * x to a dense vector with the appropriate function | 18:12 |
| -!- aifargonos [~aifargono@46.18.27.4] has joined #shogun | 18:12 | |
| jabbok | ok. thanks | 18:12 |
| blackburn | sonney2k: seems like number of applicants doubles rough estimation of slots number :) | 18:12 |
| @sonney2k | dadkins, you could for example think of doing clustering algorithms... | 18:13 |
| blackburn | for example graph clustering | 18:13 |
| dadkins | like a k-means or isodata? | 18:14 |
| blackburn | k-means is already implemented | 18:14 |
| @sonney2k | but look what is there already! | 18:14 |
| * sonney2k is off to food | 18:14 | |
| dadkins | just drawing a parallel to make sure i understood | 18:15 |
| blackburn | dadkins: isodata isn't in shogun yet | 18:16 |
| dadkins | iirc isodata is just k-means with a variable value for the number of clusters which grows and expands...that seems like a useful tool | 18:16 |
| dadkins | I also noticed that there is no self-organizing map available -- would this be useful? | 18:19 |
| jabbok | is there any way of debugging shogun besides SG_PRINT, something like building with debug symbols and afterwards using gdb ? | 18:20 |
| jabbok | I haven't seen anything in the tutorials | 18:21 |
| blackburn | dadkins: seems like there is no artificial neural networks in shogun, so SOM isn't there too | 18:21 |
| dadkins | might be best to start with a basic bpnn and go from there then... | 18:22 |
| blackburn | dadkins: don't know, try to wait for Sören, he could say his opinion on that | 18:23 |
| dadkins | blackburn: Thanks for the thoughts. The closest I could find was CPerceptron, which inherits CLinearClassifier, so my guess (without reading code) is that it won't do anything with multiple layer sigmoids...*grin* | 18:24 |
| -!- sploving [~root@124.16.139.196] has quit [Read error: Operation timed out] | 18:30 | |
| -!- sploving [~root@124.16.139.196] has joined #shogun | 18:30 | |
| -!- dadkins [~kvirc@cpe-174-102-36-192.woh.res.rr.com] has quit [Quit: KVIrc 4.0.4 Insomnia http://www.kvirc.net/] | 19:08 | |
| -!- aifargonos [~aifargono@46.18.27.4] has quit [Ping timeout: 252 seconds] | 20:02 | |
| -!- Tanmoy [75d35896@gateway/web/freenode/ip.117.211.88.150] has quit [Ping timeout: 252 seconds] | 20:28 | |
| -!- serialhex [~quassel@99-101-149-136.lightspeed.wepbfl.sbcglobal.net] has joined #shogun | 20:41 | |
| CIA-30 | shogun: Soeren Sonnenburg master * rf7f3e1d / (src/ChangeLog src/.authors): remove dirks email address from changelog - http://bit.ly/g1acX8 | 20:47 |
| @sonney2k | jabbok, --enable-debug ... but shogun is always compiled with debug symbols | 20:47 |
| @sonney2k | however it helps to disable optimization... | 20:47 |
| @sonney2k | also to reduce compile times it drastically helps to use ccache | 20:48 |
| -!- alesis-novik [~alesis@188.74.87.84] has quit [Quit: I'll be back] | 20:49 | |
| -!- aifargonos [~aifargono@46.18.27.35] has joined #shogun | 20:54 | |
| @knrrrd | The KFA task is also related to dim. reduction | 20:55 |
| -!- sonney2k [~sonney2k@g225143240.adsl.alicedsl.de] has quit [Ping timeout: 246 seconds] | 21:41 | |
| -!- sonney2k [~sonney2k@g225138211.adsl.alicedsl.de] has joined #shogun | 21:41 | |
| -!- Tanmoy [75d35896@gateway/web/freenode/ip.117.211.88.150] has joined #shogun | 21:51 | |
| -!- seviyor [c1e20418@gateway/web/freenode/ip.193.226.4.24] has quit [Ping timeout: 252 seconds] | 22:21 | |
| -!- serialhex [~quassel@99-101-149-136.lightspeed.wepbfl.sbcglobal.net] has quit [Remote host closed the connection] | 22:21 | |
| -!- serialhex [~quassel@99-101-149-136.lightspeed.wepbfl.sbcglobal.net] has joined #shogun | 23:03 | |
| CIA-30 | shogun: Soeren Sonnenburg master * r3f6962a / (3 files in 2 dirs): | 23:05 |
| CIA-30 | shogun: Implement snappy compression support (http://code.google.com/p/snappy/). | 23:05 |
| CIA-30 | shogun: This code is still experimental and pending benchmarks. - http://bit.ly/e0oZwr | 23:05 |
| -!- blackburn [~qdrgsm@188.168.2.29] has left #shogun [] | 23:28 | |
| -!- nicococo [~nicococo@d75-157-184-31.bchsia.telus.net] has joined #shogun | 23:31 | |
| --- Log closed Thu Mar 31 00:00:36 2011 | ||
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!