| --- Log opened Fri Apr 06 00:00:19 2012 | ||
| -!- flxb [~cronor@g231228114.adsl.alicedsl.de] has quit [Quit: flxb] | 00:22 | |
| -!- PhilTillet_ [~Philippe@tillet-p42154.maisel.int-evry.fr] has quit [Ping timeout: 246 seconds] | 00:41 | |
| shogun-buildbot | build #439 of ruby_modular is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/ruby_modular/builds/439 | 00:58 |
|---|---|---|
| shogun-buildbot | build #445 of python_modular is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org/buildbot/builders/python_modular/builds/445 blamelist: sonne@debian.org, gsomix@gmail.com | 01:35 |
| -!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking] | 01:41 | |
| -!- harshit_ [~harshit@182.68.94.221] has quit [Ping timeout: 252 seconds] | 01:46 | |
| -!- blackburn [~qdrgsm@83.234.54.186] has quit [Ping timeout: 246 seconds] | 02:00 | |
| -!- pluskid [~chatzilla@111.120.18.89] has joined #shogun | 03:38 | |
| -!- pluskid [~chatzilla@111.120.18.89] has quit [Client Quit] | 03:40 | |
| -!- pluskid [~chatzilla@111.120.18.89] has joined #shogun | 03:40 | |
| -!- pluskid [~chatzilla@111.120.18.89] has left #shogun [] | 03:41 | |
| -!- pluskid [~chatzilla@111.120.18.89] has joined #shogun | 03:41 | |
| -!- pluskid [~chatzilla@111.120.18.89] has quit [Client Quit] | 03:42 | |
| -!- pluskid [~chatzilla@111.120.18.89] has joined #shogun | 03:46 | |
| -!- pluskid [~chatzilla@111.120.18.89] has quit [Ping timeout: 245 seconds] | 04:25 | |
| -!- pluskid [~chatzilla@173.254.214.60] has joined #shogun | 04:25 | |
| -!- pythonroar [d2198538@gateway/web/freenode/ip.210.25.133.56] has joined #shogun | 04:46 | |
| -!- pluskid [~chatzilla@173.254.214.60] has quit [Ping timeout: 246 seconds] | 04:49 | |
| gsomix | hmm | 05:09 |
| gsomix | oops, there is wrong 'if' statement. =__= | 05:28 |
| -!- gsomix [~gsomix@188.168.3.37] has quit [Read error: Operation timed out] | 05:37 | |
| -!- KMH [~KMH@c-24-118-111-180.hsd1.mn.comcast.net] has joined #shogun | 07:01 | |
| -!- PhilTillet_ [~Philippe@tillet-p42154.maisel.int-evry.fr] has joined #shogun | 07:03 | |
| -!- KMH [~KMH@c-24-118-111-180.hsd1.mn.comcast.net] has left #shogun [] | 07:18 | |
| -!- pluskid [~chatzilla@li400-235.members.linode.com] has joined #shogun | 07:22 | |
| -!- n4nd0 [~nando@h121n3-ksg-a11.ias.bredband.telia.com] has joined #shogun | 07:31 | |
| n4nd0 | pluskid: hey! how is it going? | 07:41 |
| -!- gsomix [~gsomix@85.26.235.5] has joined #shogun | 07:44 | |
| -!- PhilTillet_ [~Philippe@tillet-p42154.maisel.int-evry.fr] has quit [Remote host closed the connection] | 07:47 | |
| -!- gsomix [~gsomix@85.26.235.5] has quit [Client Quit] | 07:48 | |
| pluskid | n4nd0: I'm officially graduated now | 07:48 |
| pluskid | finished all the busy stuff | 07:48 |
| n4nd0 | pluskid: oh that's really cool, congrats :) | 07:48 |
| pluskid | Thanks! | 07:49 |
| n4nd0 | is your major in computer science or already sth more particular? | 07:49 |
| pluskid | Now I started to read some related papers | 07:49 |
| pluskid | CS :) | 07:49 |
| pluskid | how about you? | 07:49 |
| n4nd0 | I am fine | 07:50 |
| n4nd0 | having my personal battle with cover trees for KNN :P | 07:51 |
| pluskid | what is a cover tree? | 07:51 |
| n4nd0 | is a tree structure that can be used for fast nearest neighbors | 07:52 |
| pluskid | KDTree? | 07:52 |
| n4nd0 | not sure if they're the same | 07:52 |
| pluskid | OK, I see | 07:53 |
| n4nd0 | I don't think so | 07:53 |
| n4nd0 | it seems that KDTrees are binary trees right? | 07:53 |
| pluskid | yes, IIRC | 07:53 |
| n4nd0 | cover trees are not | 07:53 |
| n4nd0 | the results people have got with them for KNN are nice | 07:54 |
| pluskid | Then how it works? Do a hierarchical clustering and use that clustering tree to search? | 07:54 |
| n4nd0 | check it out if interested | 07:54 |
| n4nd0 | http://hunch.net/~beygel/cover-tree.pdf | 07:54 |
| pluskid | Ah, hunch.net is a famous site :D | 07:54 |
| n4nd0 | if you mean with hierarchical that the nodes in the tree are stored depending on their distances between each other, yes | 07:55 |
| pluskid | So you are implementing this in shogun? | 07:57 |
| n4nd0 | the cover tree is already there | 07:57 |
| n4nd0 | is ported from an open source implementation | 07:58 |
| n4nd0 | I am integrating it with KNN | 07:58 |
| pluskid | OK, I see, you are using this to speed up KNN | 07:58 |
| n4nd0 | yeah | 07:58 |
| n4nd0 | I have it integrated already, just polishing details | 07:58 |
| pluskid | Cool! | 07:58 |
| n4nd0 | but it would be nice to have examples for speedup like the ones in the paper | 07:58 |
| pluskid | Yes, some benchmark test on some standard dataset | 07:59 |
| pluskid | Cover tree looks cool. the limitation of KDTree is that it becomes slow when the dimension is high | 08:00 |
| n4nd0 | I see, are they use for speedup as well? | 08:01 |
| pluskid | yes | 08:01 |
| pluskid | For speed up NN Search | 08:01 |
| pluskid | or do approximate NN search | 08:01 |
| pluskid | (not sure whether cover tree can do approximate NN search) | 08:02 |
| n4nd0 | I have seen cool figures related to kd trees, but didn't really know about them http://www.google.se/imgres?um=1&hl=sv&biw=1366&bih=654&tbm=isch&tbnid=oAvpRgWrcLE9VM:&imgrefurl=http://docs.pointclouds.org/trunk/group__kdtree.html&docid=vLhvyDzQfouQKM&imgurl=http://www.pointclouds.org/assets/images/contents/documentation/kdtree_mug.png&w=525&h=465&ei=v4Z-T8H3I8eM4gTutODsBw&zoom=1&iact=hc&vpx=182&vpy=172&dur=1633&hovh=211&hovw=239&tx=88&ty | 08:02 |
| pluskid | basically, it split the searching for neighbors along each dimension | 08:03 |
| pluskid | http://en.wikipedia.org/wiki/K-d_tree | 08:03 |
| pluskid | The 2D example is good for understanding | 08:04 |
| pluskid | http://upload.wikimedia.org/wikipedia/commons/thumb/b/bf/Kdtree_2d.svg/300px-Kdtree_2d.svg.png | 08:04 |
| pluskid | But when the data dimension grows, the depth of the tree grows, and searching becomes slow | 08:04 |
| n4nd0 | oh wait, yeah I did hear about this kd-trees in a course :P | 08:05 |
| pluskid | :) | 08:05 |
| pluskid | I think kd tree is widely used in practice | 08:06 |
| pluskid | another related interesting (more theory or algorithm oriented) work is random projection tree | 08:06 |
| pluskid | by Sanjoy Dusgupta: http://cseweb.ucsd.edu/~dasgupta/papers/rptree-stoc.pdf | 08:06 |
| pluskid | IIRC, it has the advantage over KDTree that | 08:07 |
| pluskid | when the *intrinsic dimension* of the data is low, its performance do not degenerate even when the extrinsic data dimension is very high (like KDTree) | 08:08 |
| n4nd0 | aham, interesting | 08:08 |
| pluskid | but I do not know whether it is already used in some real-world projects already | 08:08 |
| pluskid | n4nd0: have you worked on structured learning before? | 08:11 |
| n4nd0 | pluskid: no, just the stuff I have been reading when I got to know the project in shogun | 08:11 |
| pluskid | me too | 08:11 |
| n4nd0 | it looks really nice though | 08:11 |
| pluskid | yes, got applied in lots of practical problems | 08:12 |
| n4nd0 | you are mainly interested in computer vision, aren't you? | 08:12 |
| pluskid | actually, I heard someone else in my lab is using structured SVM to do human detection | 08:12 |
| pluskid | actually, I'm mainly in machine learning :D | 08:13 |
| n4nd0 | ok | 08:13 |
| pluskid | how about you, what's your interest? | 08:13 |
| n4nd0 | robotics | 08:13 |
| pluskid | that's cool! | 08:13 |
| n4nd0 | my master is on that | 08:13 |
| pluskid | what kind of robots do you work with? | 08:14 |
| n4nd0 | we don't get that much in contact with real robots to tell the truth :S | 08:14 |
| n4nd0 | but my focus is on autonomous systems | 08:15 |
| pluskid | so, the software part? | 08:15 |
| n4nd0 | I am interested in localization, mapping, navigation | 08:15 |
| n4nd0 | yeah sure | 08:15 |
| n4nd0 | I am a CS guy :) | 08:15 |
| pluskid | Aha! :D | 08:15 |
| pluskid | I have some very limited knowledge of reinforcement learning | 08:16 |
| pluskid | do you use RL in your work? | 08:16 |
| n4nd0 | I would like to learn more about it | 08:17 |
| n4nd0 | SO is applied in RL according to what I have heard | 08:17 |
| pluskid | oh? | 08:18 |
| pluskid | for example? | 08:18 |
| n4nd0 | for navigation | 08:19 |
| n4nd0 | like in order to optimize the best sequence of actions | 08:19 |
| n4nd0 | but I have not found more stuff beyond mentions like that one | 08:20 |
| pluskid | ok, I see | 08:20 |
| n4nd0 | pluskid: thinking of doing a new patch for shogun or already working on sth? | 08:30 |
| pluskid | reading papers | 08:30 |
| n4nd0 | ok | 08:30 |
| pluskid | you have done a lot of improvements to shogun these days! :D | 08:32 |
| n4nd0 | not that many | 08:32 |
| pluskid | I'm trying to figure out some basic stuffs related to SO implementations | 08:34 |
| pluskid | and what is already there in shogun | 08:34 |
| n4nd0 | I see | 08:34 |
| -!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 08:53 | |
| -!- uricamic [9320543b@gateway/web/freenode/ip.147.32.84.59] has joined #shogun | 08:55 | |
| -!- blackburn [~qdrgsm@83.234.54.186] has joined #shogun | 08:56 | |
| -!- n4nd0 [~nando@h121n3-ksg-a11.ias.bredband.telia.com] has quit [Quit: leaving] | 08:59 | |
| -!- wizneel [0e8be20b@gateway/web/freenode/ip.14.139.226.11] has joined #shogun | 09:06 | |
| -!- gsomix [~gsomix@188.168.13.23] has joined #shogun | 09:08 | |
| -!- blackburn [~qdrgsm@83.234.54.186] has quit [Ping timeout: 265 seconds] | 09:10 | |
| -!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking] | 09:24 | |
| wizneel | any ideas for how to integrate c5.0 as shogun macchine | 09:27 |
| -!- harshit_ [~harshit@182.68.94.221] has joined #shogun | 09:27 | |
| -!- wiking [~wiking@78-23-191-201.access.telenet.be] has joined #shogun | 09:36 | |
| -!- wiking [~wiking@78-23-191-201.access.telenet.be] has quit [Changing host] | 09:36 | |
| -!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 09:36 | |
| wizneel | hi | 09:39 |
| wizneel | any body here going for integrating c5 | 09:39 |
| pluskid | there seems to be lots of discussions related to C5.0 on the mailing list | 09:40 |
| -!- KMH [~KMH@c-24-118-111-180.hsd1.mn.comcast.net] has joined #shogun | 09:41 | |
| KMH | I am almost done with my application except the "set yourself up" part. I downloaded shogun using git in my cygwin. | 09:42 |
| -!- wizneel [0e8be20b@gateway/web/freenode/ip.14.139.226.11] has quit [Ping timeout: 245 seconds] | 09:46 | |
| -!- muddo [~muddo@gateway/tor-sasl/muddo] has quit [Ping timeout: 276 seconds] | 09:50 | |
| -!- muddo [~muddo@gateway/tor-sasl/muddo] has joined #shogun | 10:02 | |
| -!- muddo [~muddo@gateway/tor-sasl/muddo] has quit [Ping timeout: 276 seconds] | 10:08 | |
| -!- PhilTillet [9d9f2503@gateway/web/freenode/ip.157.159.37.3] has joined #shogun | 10:16 | |
| PhilTillet | Hey :) | 10:17 |
| -!- wizneel [0e8be20b@gateway/web/freenode/ip.14.139.226.11] has joined #shogun | 10:17 | |
| -!- muddo [~muddo@gateway/tor-sasl/muddo] has joined #shogun | 10:20 | |
| KMH | I set up most of it | 10:24 |
| KMH | but I get shogun/base/init.h not found | 10:24 |
| -!- muddo [~muddo@gateway/tor-sasl/muddo] has quit [Ping timeout: 276 seconds] | 10:28 | |
| -!- muddo [~muddo@gateway/tor-sasl/muddo] has joined #shogun | 10:39 | |
| -!- nickon [~noneedtok@dD5774105.access.telenet.be] has joined #shogun | 10:52 | |
| -!- emrecelikten [~Anubis@176.40.235.143] has joined #shogun | 10:58 | |
| emrecelikten | Hello all! | 11:00 |
| -!- harshit_ [~harshit@182.68.94.221] has quit [Ping timeout: 240 seconds] | 11:00 | |
| -!- PhilTillet [9d9f2503@gateway/web/freenode/ip.157.159.37.3] has quit [Quit: Page closed] | 11:00 | |
| -!- harshit_ [~harshit@182.68.94.221] has joined #shogun | 11:01 | |
| -!- PhilTillet [9d9f2503@gateway/web/freenode/ip.157.159.37.3] has joined #shogun | 11:10 | |
| -!- KMH [~KMH@c-24-118-111-180.hsd1.mn.comcast.net] has quit [Ping timeout: 252 seconds] | 11:23 | |
| -!- PhilTillet [9d9f2503@gateway/web/freenode/ip.157.159.37.3] has quit [Ping timeout: 245 seconds] | 11:25 | |
| -!- pythonroar [d2198538@gateway/web/freenode/ip.210.25.133.56] has quit [Ping timeout: 245 seconds] | 11:26 | |
| -!- harshit_ [~harshit@182.68.94.221] has quit [Quit: Leaving] | 11:52 | |
| -!- wizneel [0e8be20b@gateway/web/freenode/ip.14.139.226.11] has quit [Ping timeout: 245 seconds] | 11:56 | |
| -!- n4nd0 [~nando@h121n3-ksg-a11.ias.bredband.telia.com] has joined #shogun | 12:00 | |
| n4nd0 | sonney2k: I hope there is no problem with my answer in the mailing list about the "patch's deadline" | 12:07 |
| emrecelikten | Oh so you're Fernando | 12:09 |
| emrecelikten | :P | 12:09 |
| emrecelikten | Well that's good to know if it is as you say. | 12:09 |
| n4nd0 | I remember that someone already asked around here in IRC about a deadline for a patch | 12:10 |
| n4nd0 | I just transmitted that answer ;) | 12:11 |
| emrecelikten | Thanks :) | 12:12 |
| n4nd0 | no problem | 12:14 |
| n4nd0 | wiking: around? | 12:19 |
| -!- Afan_ [~Afan@c-76-29-50-209.hsd1.il.comcast.net] has joined #shogun | 12:20 | |
| -!- n4nd0 [~nando@h121n3-ksg-a11.ias.bredband.telia.com] has quit [Quit: leaving] | 12:29 | |
| -!- karlnapf [~heiko@host86-177-14-82.range86-177.btcentralplus.com] has joined #shogun | 12:31 | |
| -!- harshit_ [~harshit@182.68.94.221] has joined #shogun | 12:32 | |
| gsomix | hi | 12:35 |
| wiking | yes | 12:36 |
| wiking | n4nd0 | 12:36 |
| -!- harshit_ [~harshit@182.68.94.221] has quit [Ping timeout: 240 seconds] | 12:43 | |
| -!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has joined #shogun | 12:47 | |
| -!- n4nd0 [~androirc@h121n3-ksg-a11.ias.bredband.telia.com] has joined #shogun | 12:47 | |
| gsomix | blackburn, yo | 12:47 |
| blackburn | hey there | 12:47 |
| -!- harshit_ [~harshit@182.68.94.221] has joined #shogun | 12:49 | |
| karlnapf | sonney2k, blackburn, around? | 12:57 |
| blackburn | karlnapf: yeap | 12:57 |
| karlnapf | Just working on subsets and thinking of the case when the subset is larger than the number of features | 12:57 |
| karlnapf | should we allow this? | 12:57 |
| karlnapf | then people could kind of "duplicate" features | 12:57 |
| karlnapf | but I dont know if that might kill something else | 12:58 |
| blackburn | something like | 12:58 |
| blackburn | 1 2 1? | 12:58 |
| karlnapf | yes | 12:59 |
| blackburn | do you think it makes sense? | 12:59 |
| karlnapf | if you want to have duplicate features for some reason | 13:00 |
| karlnapf | mmh | 13:00 |
| karlnapf | no it makes no sense i guess | 13:00 |
| karlnapf | :D | 13:00 |
| blackburn | if there was at least one use case of that.. | 13:00 |
| blackburn | for training NNs probably | 13:01 |
| karlnapf | MMh I think I will put an error message and if people complain I will allow it | 13:02 |
| blackburn | yeah sure | 13:03 |
| blackburn | they didn't yet, right? | 13:04 |
| blackburn | :) | 13:04 |
| -!- harshit_ [~harshit@182.68.94.221] has quit [Remote host closed the connection] | 13:05 | |
| karlnapf | hehe :) | 13:05 |
| karlnapf | indeed | 13:05 |
| -!- iglesias [~androirc@h121n3-ksg-a11.ias.bredband.telia.com] has joined #shogun | 13:06 | |
| -!- n4nd0 [~androirc@h121n3-ksg-a11.ias.bredband.telia.com] has quit [Read error: Connection reset by peer] | 13:06 | |
| pluskid | repetition sometimes can be used as poor man's way of doing re-weighting of samples | 13:14 |
| pluskid | but not all algorithm can work with duplicated data | 13:15 |
| pluskid | for re-weighting, I mean, for example, you have 1000 positive samples but only 10 negative samples | 13:18 |
| blackburn | yeah but it is rather poor | 13:18 |
| pluskid | sure | 13:20 |
| @sonney2k | karlnapf, don't try to solve a problem we don't have yet :D | 13:23 |
| @sonney2k | pluskid, yeah I used that too :D | 13:23 |
| pluskid | haha | 13:23 |
| karlnapf | sonney2k, yeah, sounds like a good idea :) | 13:23 |
| karlnapf | pluskid, I also did that at some -late night bugfixing and not knowing wheres the error-point | 13:24 |
| karlnapf | :) | 13:24 |
| pluskid | you mean duplication of data was causing errors? :p | 13:25 |
| karlnapf | no I was searching why my methos did not work and duplicated data to check if the results were consistent | 13:27 |
| pluskid | Ah, I see | 13:27 |
| karlnapf | so it finally helped my finding the error ;) | 13:27 |
| pluskid | cool | 13:27 |
| karlnapf | but this is all not how it should be I guess | 13:27 |
| pluskid | maybe only work for some particular case/algor | 13:28 |
| -!- iglesias [~androirc@h121n3-ksg-a11.ias.bredband.telia.com] has quit [Quit: AndroIRC - Android IRC Client ( http://www.androirc.com )] | 13:28 | |
| -!- karlnapf1 [~heiko@host86-181-10-46.range86-181.btcentralplus.com] has joined #shogun | 13:32 | |
| blackburn | sonney2k: what is # of proposals btw? | 13:32 |
| -!- karlnapf [~heiko@host86-177-14-82.range86-177.btcentralplus.com] has quit [Ping timeout: 276 seconds] | 13:34 | |
| @sonney2k | maybe 35 | 13:38 |
| -!- vikram360 [~vikram360@117.192.160.138] has joined #shogun | 13:38 | |
| @sonney2k | no idea | 13:38 |
| pluskid | that's a big number! | 13:39 |
| pluskid | the quality and quantity of proposals will affect how many slots you get, I guess | 13:39 |
| -!- PhilTillet [~Philippe@npasserelle10.minet.net] has joined #shogun | 13:46 | |
| @sonney2k | 38 | 13:49 |
| blackburn | huh | 13:49 |
| PhilTillet | hi sonney2k :) | 13:49 |
| @sonney2k | blackburn, I get about 8 emails about updated/new proposals per hour now... | 13:50 |
| blackburn | sonney2k: yeah I guess it will be 16 this evening | 13:51 |
| blackburn | I have to update my proposal as well | 13:51 |
| blackburn | sonney2k: things stay the same, right? you said last year there were a lot of proposal before deadline | 13:53 |
| emrecelikten | sonney2k: Sorry for the spam :/ | 13:53 |
| -!- n4nd0 [~nando@h121n3-ksg-a11.ias.bredband.telia.com] has joined #shogun | 13:53 | |
| emrecelikten | 38 proposals... So, the probability of being accepted for a candidate is roughly 0.1316 then | 13:54 |
| emrecelikten | assuming 5 slots | 13:54 |
| emrecelikten | and equal chance for everybody | 13:54 |
| n4nd0 | 38 proposals! that's a good number I guess | 13:55 |
| wiking | n4nd0: you were looking for me | 13:55 |
| n4nd0 | wiking: I needed git support a moment | 13:56 |
| wiking | n4nd0: sure shoot | 13:56 |
| n4nd0 | wiking: I did a pull request without rebasing master, later I rebase in my master and in my branch but when I wanted to push I got the error of non-fast forward allowed | 13:56 |
| n4nd0 | wiking: they suggested me to pull | 13:57 |
| n4nd0 | wiking: I asked blackburn and he told me that pulling was the best option probably | 13:57 |
| -!- karlnapf1 [~heiko@host86-181-10-46.range86-181.btcentralplus.com] has quit [Ping timeout: 276 seconds] | 13:57 | |
| wiking | n4nd0: yep you'll need to do a pull first in order to do the sync with your remote repo on github | 13:58 |
| n4nd0 | wiking: idk if that would make a problem related to the pull request though, like sending commits done by other people | 13:58 |
| wiking | well | 13:58 |
| wiking | it won't include the other commits | 13:58 |
| @sonney2k | last year we had more than 70 though | 13:58 |
| wiking | of other people.... just yours and the merging | 13:58 |
| n4nd0 | wiking: I am going to try now, let's see | 13:59 |
| pluskid | n4nd0: delete your remote branch and push again | 14:03 |
| n4nd0 | pluskid: no need to do that :) | 14:03 |
| emrecelikten | I'm looking for ideas to make a patch about. Suggestions are welcome :) | 14:04 |
| n4nd0 | emrecelikten: yes | 14:05 |
| n4nd0 | emrecelikten: so I have recently added CoverTree support for KNN | 14:05 |
| n4nd0 | emrecelikten: a CoverTree if just an structure that allows to do faster k-neares neighbors, just in case you didn't hear about it (it was new for me too a couple of weeks ago) | 14:05 |
| n4nd0 | emrecelikten: I would like to see some examples of benchmarking, that shows that CoverTree is actually faster than the KNN we had before | 14:06 |
| n4nd0 | emrecelikten: http://hunch.net/~jl/projects/cover_tree/cover_tree.html | 14:06 |
| n4nd0 | emrecelikten: something like that ^ | 14:06 |
| n4nd0 | blackburn, sonney2k: do you think guys that could be a good idea for a patch? | 14:07 |
| blackburn | yes | 14:07 |
| blackburn | n4nd0: I admit you become as lazy as me haha | 14:07 |
| n4nd0 | blackburn: haha :P I was planning to start with LARS or other thing then ;) | 14:07 |
| blackburn | ah yeah | 14:07 |
| blackburn | n4nd0: btw probably I can suggest you some other way of obtaining lasso pathwise solution | 14:08 |
| n4nd0 | blackburn: tell me | 14:09 |
| blackburn | from the SLEP that was referred by pluskid before | 14:09 |
| n4nd0 | blackburn: ok, I took a look to the page | 14:09 |
| pluskid | n4nd0: so you solved with a git pull? :) | 14:09 |
| n4nd0 | blackburn: which one do you like? | 14:09 |
| emrecelikten | Okay, thanks. So I need to make some benchmarking tests then? How am I going to do it? Sample Octave programs? | 14:09 |
| n4nd0 | emrecelikten: the good thing here is that you may have freedom to select the language you want to use, if you like octave, octave would be cool | 14:10 |
| n4nd0 | pluskid: yeah, it add an empty merge but fixed it :) | 14:10 |
| emrecelikten | Okay, good | 14:10 |
| emrecelikten | n4nd0: Thanks! :) | 14:10 |
| n4nd0 | the only issue is that this is not yet in the central repository | 14:10 |
| blackburn | n4nd0: actually group/tree/etc lasso are my task | 14:10 |
| n4nd0 | I have submitted the pull request this morning | 14:11 |
| blackburn | n4nd0: but you can do simple L1 | 14:11 |
| n4nd0 | blackburn: how could we fix this? so emrecelikten can do some benchmarking with cover tree even if it is not yet merged? | 14:11 |
| blackburn | n4nd0: he can git pull git://github.com/iglesias/shogun yourbranchwithcovertree | 14:11 |
| * emrecelikten is taking notes | 14:12 | |
| n4nd0 | emrecelikten: then that is probably the best to do | 14:12 |
| pluskid | n4nd0: cool. I also got similar problem, I solved by deleting the remote branch on my github and push the new local branch there again -- an alternative solution :p | 14:13 |
| n4nd0 | pluskid: good to know :) | 14:13 |
| n4nd0 | emrecelikten: you can create a new branch locally and pull mine's | 14:13 |
| emrecelikten | Okay, I will do this then. Maybe tonight or tomorrow, it will definitely be done until Sunday | 14:14 |
| n4nd0 | emrecelikten: ok, probably to use a famous dataset like iris, http://archive.ics.uci.edu/ml/datasets/Iris, for the benchmarking would be interesting | 14:15 |
| @sonney2k | n4nd0, look at mldata.org | 14:15 |
| @sonney2k | :D | 14:15 |
| emrecelikten | Will do | 14:16 |
| @sonney2k | <ad>enough datasets all in a format easy to read </ad> | 14:16 |
| -!- Afan_ [~Afan@c-76-29-50-209.hsd1.il.comcast.net] has quit [Read error: Connection reset by peer] | 14:17 | |
| n4nd0 | sonney2k: thanks! | 14:19 |
| -!- wizneel [0e8be20b@gateway/web/freenode/ip.14.139.226.11] has joined #shogun | 14:44 | |
| -!- PhilTillet [~Philippe@npasserelle10.minet.net] has quit [Ping timeout: 245 seconds] | 14:48 | |
| -!- PhilTillet [~Philippe@npasserelle10.minet.net] has joined #shogun | 15:00 | |
| -!- muddo [~muddo@gateway/tor-sasl/muddo] has quit [Remote host closed the connection] | 15:10 | |
| -!- muddo [~muddo@gateway/tor-sasl/muddo] has joined #shogun | 15:10 | |
| -!- vikram360 [~vikram360@117.192.160.138] has quit [Ping timeout: 265 seconds] | 15:28 | |
| -!- uricamic [9320543b@gateway/web/freenode/ip.147.32.84.59] has quit [Quit: Page closed] | 15:28 | |
| -!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has quit [Quit: Page closed] | 15:34 | |
| PhilTillet | I am tired :p | 15:35 |
| n4nd0 | PhilTillet: of? | 15:36 |
| PhilTillet | of nothing | 15:37 |
| PhilTillet | I am just exhausted | 15:37 |
| PhilTillet | :p | 15:37 |
| PhilTillet | sleepu | 15:37 |
| PhilTillet | sleepy* | 15:37 |
| n4nd0 | aham I see | 15:38 |
| n4nd0 | see you later guys | 15:39 |
| -!- n4nd0 [~nando@h121n3-ksg-a11.ias.bredband.telia.com] has quit [Quit: leaving] | 15:39 | |
| -!- wizneel [0e8be20b@gateway/web/freenode/ip.14.139.226.11] has quit [Ping timeout: 245 seconds] | 15:58 | |
| -!- wizneel [7375d3a1@gateway/web/freenode/ip.115.117.211.161] has joined #shogun | 16:07 | |
| gsomix | sonney2k, hello | 16:09 |
| -!- n4nd0 [~nando@h121n3-ksg-a11.ias.bredband.telia.com] has joined #shogun | 16:10 | |
| -!- wizneel [7375d3a1@gateway/web/freenode/ip.115.117.211.161] has quit [Quit: Page closed] | 16:15 | |
| gsomix | I completed proposal. | 16:25 |
| gsomix | and I'm happy | 16:26 |
| n4nd0 | gsomix: :) | 16:27 |
| -!- wizneel_ [0e8be20b@gateway/web/freenode/ip.14.139.226.11] has joined #shogun | 16:35 | |
| emrecelikten | Here comes another proposal update email :P | 16:36 |
| -!- wizneel_ [0e8be20b@gateway/web/freenode/ip.14.139.226.11] has quit [Client Quit] | 16:37 | |
| -!- genix [~gsomix@188.168.4.75] has joined #shogun | 16:39 | |
| -!- gsomix [~gsomix@188.168.13.23] has quit [Ping timeout: 260 seconds] | 16:42 | |
| -!- blackburn [~qdrgsm@83.234.54.186] has joined #shogun | 16:50 | |
| -!- pythonroar_ [d2198538@gateway/web/freenode/ip.210.25.133.56] has joined #shogun | 16:51 | |
| emrecelikten | I'm curious about the amount of Dunning-Kruger effect in this programming skills self-assessment thing in proposals | 16:54 |
| emrecelikten | Especially in Shogun's 1-5 scale | 16:55 |
| blackburn | I put a 4 to shogun actually | 16:55 |
| blackburn | can it be an artifact of that effect? :) | 16:55 |
| emrecelikten | Depends :P | 16:56 |
| emrecelikten | but seriously, I know people that could put 4 to their C++ skills although they have used it less than half the amount | 16:58 |
| emrecelikten | I said 2 | 16:58 |
| emrecelikten | half the amount that I used it* | 16:59 |
| n4nd0 | that's probably why it is always better to show one's skills and let others assess them | 17:02 |
| n4nd0 | at least in my opinion of course ;) | 17:03 |
| emrecelikten | That's true, but if a student has no way of/time for showing it and the organization thinks that self-assessed scores are important then it's risky | 17:06 |
| blackburn | I don't think it is the case | 17:06 |
| emrecelikten | You might end up with a Dunning-Kruger'ed 2 while rejecting a Dunning-Kruger'ed 4 :P | 17:06 |
| emrecelikten | Well not in here, of course | 17:06 |
| emrecelikten | since you have to submit a patch :) | 17:06 |
| blackburn | we have a lot of students there that showed their skills :) | 17:06 |
| * n4nd0 is looking up what Dunning-Kruger means ... | 17:07 | |
| -!- harshit_ [~harshit@182.68.94.221] has joined #shogun | 17:07 | |
| emrecelikten | n4nd0: It's a cognitive bias | 17:07 |
| emrecelikten | n4nd0: Unskilled people over-estimate their abilities while skilled people do the exact opposite | 17:07 |
| emrecelikten | It's really interesting | 17:08 |
| -!- harshit_ [~harshit@182.68.94.221] has quit [Remote host closed the connection] | 17:22 | |
| -!- blackburn [~qdrgsm@83.234.54.186] has quit [Quit: Leaving.] | 17:23 | |
| -!- nickon [~noneedtok@dD5774105.access.telenet.be] has quit [Quit: ( www.nnscript.com :: NoNameScript 4.22 :: www.esnation.com )] | 17:28 | |
| genix | but the converse is not true. | 17:31 |
| -!- vikram360 [~vikram360@117.192.160.138] has joined #shogun | 17:31 | |
| emrecelikten | genix: How? It says that "the miscalibration of the incompetent stems from an error about the self, whereas the miscalibration of the highly competent stems from an error about others" | 17:33 |
| -!- blackburn [5bde8018@gateway/web/freenode/ip.91.222.128.24] has joined #shogun | 17:45 | |
| -!- pythonroar_ [d2198538@gateway/web/freenode/ip.210.25.133.56] has quit [Quit: Page closed] | 17:56 | |
| pluskid | crafted a new proposal (so one more spam to the inbox of all the mentors) :p | 18:11 |
| pluskid | time for sleep here, g9t all | 18:11 |
| -!- pluskid [~chatzilla@li400-235.members.linode.com] has quit [Quit: ChatZilla 0.9.88.1 [Firefox 11.0/20120312181643]] | 18:14 | |
| n4nd0 | good night or whatever it's there :P | 18:19 |
| -!- n4nd0 [~nando@h121n3-ksg-a11.ias.bredband.telia.com] has quit [Ping timeout: 248 seconds] | 18:24 | |
| -!- Kshiteesh [~KMH@c-24-118-111-180.hsd1.mn.comcast.net] has joined #shogun | 18:43 | |
| PhilTillet | sonney2k, did you have some time to try my patch? | 18:49 |
| blackburn | sonney2k: libsvm was released on April,1st, I will diff it and update accordingly | 19:04 |
| blackburn | sonney2k: nothing changed | 19:19 |
| -!- blackburn [5bde8018@gateway/web/freenode/ip.91.222.128.24] has quit [Ping timeout: 245 seconds] | 19:37 | |
| @sonney2k | alright then 47 proposals... | 19:44 |
| -!- vikram360 [~vikram360@117.192.160.138] has quit [Ping timeout: 260 seconds] | 19:46 | |
| -!- blackburn [5bde8018@gateway/web/freenode/ip.91.222.128.24] has joined #shogun | 19:56 | |
| -!- blackburn [5bde8018@gateway/web/freenode/ip.91.222.128.24] has quit [Client Quit] | 19:58 | |
| emrecelikten | sonney2k: I had a question about C5.0 | 20:07 |
| emrecelikten | sonney2k: How long do you think that project will take? | 20:13 |
| emrecelikten | sonney2k: I don't think it will be longer than 5 weeks for a student working full time | 20:14 |
| -!- n4nd0 [~androirc@h121n3-ksg-a11.ias.bredband.telia.com] has joined #shogun | 20:14 | |
| @sonney2k | emrecelikten, hopefully not - I had in mind other tree implementations then | 20:15 |
| @sonney2k | e.g. port j45 / id3 ... | 20:15 |
| n4nd0 | sonney2k: I am sorry if you get mails for every update :S | 20:15 |
| @sonney2k | no worries you will be busy | 20:16 |
| @sonney2k | were just a couple of hundred today | 20:16 |
| emrecelikten | sonney2k: Very good. But I have prepared my proposal assuming that I will only be working on C5.0 for 10 weeks, I hope that's okay? I included that I thought it would be over much sooner and I would be able to take additional tasks in it though | 20:18 |
| * genix have returned from dance floor. Again. | 20:22 | |
| -!- genix is now known as gsomix | 20:22 | |
| -!- n4nd0_ [~nando@h121n3-ksg-a11.ias.bredband.telia.com] has joined #shogun | 20:55 | |
| -!- n4nd0_ [~nando@h121n3-ksg-a11.ias.bredband.telia.com] has quit [Client Quit] | 20:58 | |
| n4nd0 | So I think that the deadline is over now! | 21:02 |
| wiking | :> | 21:03 |
| wiking | yeps | 21:03 |
| n4nd0 | How many proposals at the end? | 21:03 |
| n4nd0 | Did we reach 50? | 21:03 |
| -!- vikram360 [~vikram360@117.192.160.138] has joined #shogun | 21:07 | |
| -!- karlnapf [~heiko@host86-185-9-173.range86-185.btcentralplus.com] has joined #shogun | 21:08 | |
| PhilTillet | yes, deadline is over :p | 21:14 |
| PhilTillet | the line is dead | 21:14 |
| @sonney2k | final number is 48 | 21:17 |
| -!- gsomix [~gsomix@188.168.4.75] has quit [Ping timeout: 252 seconds] | 21:18 | |
| @sonney2k | from 38 students | 21:19 |
| @sonney2k | last years acceptance rate of students was rather high (almost 30% http://google-opensource.blogspot.de/2011/05/google-summer-of-code-where-are.html ) | 21:21 |
| n4nd0 | Pie charts rock! | 21:26 |
| PhilTillet | sonney2k, is there any other patch I can work on until next week, or are you already overwhelmed with patches? :p | 21:29 |
| -!- gsomix [~gsomix@188.168.5.2] has joined #shogun | 21:33 | |
| CIA-64 | shogun: Evgeniy Andreev master * r3e38b9e / src/interfaces/python_modular/swig_typemaps.i : fixed stupid bug with 'if' statement - http://git.io/vZXbpw | 21:39 |
| CIA-64 | shogun: Evgeniy Andreev master * r708770a / (171 files in 2 dirs): fixed examples and removed 'xrange' from class_list.cpp.py - http://git.io/w_27Lg | 21:39 |
| CIA-64 | shogun: Soeren Sonnenburg master * raa97c39 / (172 files in 3 dirs): | 21:39 |
| CIA-64 | shogun: Merge pull request #427 from gsomix/python3_interface | 21:39 |
| CIA-64 | shogun: Convert python examples to be python3 compatible - http://git.io/9YK8Vg | 21:39 |
| gsomix | yeah | 21:39 |
| gsomix | shogun-buildbot, I rely on you. | 21:41 |
| -!- vikram360 [~vikram360@117.192.160.138] has quit [Ping timeout: 265 seconds] | 21:48 | |
| -!- vikram360 [~vikram360@117.192.160.138] has joined #shogun | 21:56 | |
| shogun-buildbot | build #619 of octave_static is complete: Failure [failed git] Build details are at http://www.shogun-toolbox.org/buildbot/builders/octave_static/builds/619 blamelist: gsomix@gmail.com | 21:58 |
| -!- Kshiteesh [~KMH@c-24-118-111-180.hsd1.mn.comcast.net] has quit [] | 22:05 | |
| shogun-buildbot | build #620 of octave_static is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/octave_static/builds/620 | 22:21 |
| -!- n4nd0_ [~nando@h121n3-ksg-a11.ias.bredband.telia.com] has joined #shogun | 22:38 | |
| n4nd0_ | sonney2k: hi! I am making the fixes for KNN right now, sorry for the conventions I didn't follow! | 22:39 |
| n4nd0_ | sonney2k: apart from that, do you think it is ok? | 22:39 |
| @sonney2k | n4nd0_, as usual yes - and don't worry. | 22:44 |
| n4nd0_ | sonney2k: :) do you have any suggestion to work on after this? | 22:46 |
| n4nd0_ | sonney2k: is LARS ok or any other algorithm for lasso path? any other thing of interest? | 22:47 |
| -!- n4nd0 [~androirc@h121n3-ksg-a11.ias.bredband.telia.com] has quit [Ping timeout: 246 seconds] | 22:58 | |
| @sonney2k | n4nd0_, what would help us most is to fix the test suite | 23:00 |
| @sonney2k | n4nd0_, so run tester.py and determine the examples that don't work | 23:00 |
| -!- harshit_ [~harshit@182.68.94.221] has joined #shogun | 23:00 | |
| @sonney2k | then figure out why by running tester.py -d filename.py | 23:01 |
| @sonney2k | to enter debug mode when sth is not the same | 23:01 |
| @sonney2k | p a will print object a | 23:01 |
| @sonney2k | p b will print object b | 23:01 |
| @sonney2k | and then one can check where they differ /how | 23:02 |
| n4nd0_ | ok, I will take a look to that then | 23:04 |
| shogun-buildbot | build #446 of python_modular is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/python_modular/builds/446 | 23:04 |
| gsomix | shogun-buildbot, thanks. | 23:08 |
| n4nd0_ | good night people | 23:10 |
| @sonney2k | nite | 23:10 |
| -!- n4nd0_ [~nando@h121n3-ksg-a11.ias.bredband.telia.com] has quit [Quit: leaving] | 23:10 | |
| gsomix | sonney2k, moin. | 23:10 |
| -!- emrecelikten [~Anubis@176.40.235.143] has left #shogun [] | 23:11 | |
| gsomix | sonney2k, can I switch to work with octave modular (sparse matrix typemap)? what else is needed for the python3 interface? | 23:13 |
| @sonney2k | gsomix, I don't know - I assume all examples run again? | 23:18 |
| @sonney2k | if so go ahead and do the octave modular sparse typemap. | 23:19 |
| @sonney2k | ask if you want a hint :D | 23:19 |
| gsomix | sonney2k, hint? :] | 23:19 |
| @sonney2k | gsomix, help or so | 23:20 |
| gsomix | sonney2k, I see. I thought that you have special hint for this. | 23:22 |
| @sonney2k | btw you should also fix the python_static examples' print stuff | 23:22 |
| gsomix | sonney2k, ok. | 23:22 |
| @sonney2k | gsomix, well look at how it is done in interfaces/octave_static/OctaveInterface.cpp | 23:22 |
| @sonney2k | function COctaveInterface::get_sparse_matrix | 23:23 |
| @sonney2k | set_sparse_matrix | 23:23 |
| @sonney2k | basically copy and paste and rename this stuff and you are done | 23:23 |
| @sonney2k | ah and note that sparse matrices can only be of double type in octave | 23:24 |
| gsomix | sonney2k, tnx. you make my life easier. :) | 23:27 |
| gsomix | good night, guys | 23:36 |
| -!- harshit_ [~harshit@182.68.94.221] has quit [Remote host closed the connection] | 23:43 | |
| -!- blackburn [5bde8018@gateway/web/freenode/ip.91.222.128.24] has joined #shogun | 23:52 | |
| --- Log closed Sat Apr 07 00:00:15 2012 | ||
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!
