--- Log opened Thu Apr 12 00:00:19 2012 | ||
CIA-64 | shogun: Sergey Lisitsyn master * ra826101 / src/shogun/classifier/svm/LaRank.h : Got rid of larank deprecated warning - http://git.io/ZsDCuA | 00:02 |
---|---|---|
-!- harshit_ [~harshit@182.68.45.192] has quit [Quit: Leaving] | 00:13 | |
-!- PhilTillet [~Philippe@npasserelle10.minet.net] has joined #shogun | 00:39 | |
PhilTillet | Hey :p | 00:40 |
-!- PhilTillet [~Philippe@npasserelle10.minet.net] has quit [Ping timeout: 245 seconds] | 00:52 | |
-!- PhilTillet [~Philippe@npasserelle10.minet.net] has joined #shogun | 00:54 | |
-!- PhilTillet [~Philippe@npasserelle10.minet.net] has quit [Client Quit] | 00:55 | |
-!- PhilTillet [~Philippe@npasserelle10.minet.net] has joined #shogun | 00:55 | |
-!- PhilTillet [~Philippe@npasserelle10.minet.net] has quit [Ping timeout: 250 seconds] | 00:59 | |
-!- av3ngr [av3ngr@nat/redhat/x-bhvghgippgadxuui] has joined #shogun | 01:05 | |
-!- av3ngr [av3ngr@nat/redhat/x-bhvghgippgadxuui] has quit [Client Quit] | 01:06 | |
-!- av3ngr [av3ngr@nat/redhat/x-dwqkcbcthexeyton] has joined #shogun | 01:06 | |
-!- av3ngr [av3ngr@nat/redhat/x-dwqkcbcthexeyton] has quit [Remote host closed the connection] | 01:10 | |
-!- PhilTillet [~Philippe@157.159.42.154] has joined #shogun | 01:13 | |
-!- PhilTillet [~Philippe@157.159.42.154] has quit [Remote host closed the connection] | 01:35 | |
-!- blackburn [~qdrgsm@83.234.54.186] has quit [Quit: Leaving.] | 01:51 | |
-!- romi_ [~mizobe@187.66.121.180] has joined #shogun | 02:15 | |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking] | 02:57 | |
-!- av3ngr [av3ngr@nat/redhat/x-ynyjtavjttwlbxfx] has joined #shogun | 02:59 | |
-!- vikram360 [~vikram360@117.192.180.124] has quit [Ping timeout: 245 seconds] | 03:05 | |
-!- vikram360 [~vikram360@117.192.180.124] has joined #shogun | 04:12 | |
-!- vikram360 [~vikram360@117.192.180.124] has quit [Ping timeout: 265 seconds] | 04:26 | |
-!- vikram360 [~vikram360@117.192.181.226] has joined #shogun | 05:33 | |
@sonney2k | *drum roll* | 06:08 |
@sonney2k | 9 slots | 06:08 |
-!- n4nd0 [~androirc@s83-179-44-135.cust.tele2.se] has joined #shogun | 06:23 | |
-!- n4nd0 [~androirc@s83-179-44-135.cust.tele2.se] has quit [Remote host closed the connection] | 06:28 | |
-!- n4nd0 [~androirc@s83-179-44-135.cust.tele2.se] has joined #shogun | 06:42 | |
-!- vikram360 [~vikram360@117.192.181.226] has quit [Ping timeout: 260 seconds] | 06:44 | |
-!- vikram360 [~vikram360@117.192.171.231] has joined #shogun | 06:47 | |
n4nd0 | That's awesome! | 06:48 |
-!- n4nd0 [~androirc@s83-179-44-135.cust.tele2.se] has quit [Quit: AndroIRC - Android IRC Client ( http://www.androirc.com )] | 07:02 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 07:02 | |
-!- harshit_ [~harshit@59.177.195.46] has joined #shogun | 07:08 | |
harshit_ | n4nd0: Hola ! | 07:12 |
n4nd0 | harshit_: ey! how is it going? | 07:16 |
harshit_ | hey, my TSVM is almost ready but the problem is it takes too much time to converge | 07:17 |
harshit_ | whereas in svmlin time it doesn't takes that much time | 07:17 |
harshit_ | dont knw whats going on!! | 07:18 |
n4nd0 | harshit_: aham, so there must something different between both implementations right=? | 07:18 |
n4nd0 | mmmm about time of convergence, have you checked if they use a kind of tolerance value | 07:19 |
harshit_ | yeah, but I havnt changed that part much in shogun. that was already in shogun | 07:19 |
harshit_ | tolerance value ?? | 07:19 |
n4nd0 | yeah, it was just an idea ... | 07:19 |
harshit_ | how to check that..? | 07:19 |
n4nd0 | looking at their code | 07:19 |
n4nd0 | anyhow, are you confident then that it is not caused by the code you have written but by the other code? | 07:20 |
harshit_ | might be .! dont know exactly at this time .. have to debug a lil more | 07:21 |
harshit_ | n4nd0: Hows your covertree | 07:21 |
n4nd0 | harshit_: I have to discuss with blackburn some details before starting to do the port | 07:25 |
n4nd0 | but it looks promising, it is very fast | 07:25 |
harshit_ | great .. Maybe after completing on KNN we can apply it on K-Means , | 07:25 |
harshit_ | what sya | 07:25 |
harshit_ | say* | 07:25 |
-!- blackburn [~qdrgsm@83.234.54.186] has joined #shogun | 07:37 | |
n4nd0 | blackburn: hey! have you seen the good news?! | 07:38 |
blackburn | yeah | 07:38 |
n4nd0 | blackburn: really good :D | 07:39 |
blackburn | sure better than expectance | 07:39 |
harshit_ | n4nd0: whats the good news ? | 07:40 |
n4nd0 | harshit_: 9 slots!! | 07:42 |
harshit_ | dam awesome ! | 07:43 |
n4nd0 | yeah | 07:44 |
-!- harshit_ [~harshit@59.177.195.46] has quit [Ping timeout: 250 seconds] | 07:58 | |
-!- blackburn [~qdrgsm@83.234.54.186] has quit [Quit: Leaving.] | 08:06 | |
gsomix | hi all | 08:16 |
@sonney2k | we are currently investigating if we can handle all 9 or just 8 (as we requested) | 08:16 |
@sonney2k | s/just/'just'/ | 08:16 |
@sonney2k | it is a huuuge number | 08:16 |
gsomix | and good morning | 08:16 |
n4nd0 | gsomix: good morning | 08:16 |
n4nd0 | sonney2k: really, could it be a problem with the number of mentors? | 08:16 |
gsomix | 9? | 08:18 |
n4nd0 | gsomix: yeah | 08:19 |
gsomix | time to vodka | 08:19 |
n4nd0 | haha | 08:19 |
n4nd0 | c'mon isn't it like 10 around there? :P | 08:19 |
@sonney2k | n4nd0, yes | 08:24 |
n4nd0 | that's not good | 08:24 |
@sonney2k | not enough mentors | 08:24 |
@sonney2k | we had more ideas but for some ideas we did not get good proposals | 08:24 |
@sonney2k | so these mentors won't mentor anything | 08:24 |
@sonney2k | so in fact we are 7-8 mentors | 08:25 |
@sonney2k | alright | 08:25 |
@sonney2k | have to leave train | 08:25 |
n4nd0 | tty later | 08:25 |
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has joined #shogun | 08:35 | |
sonne|work | re | 08:41 |
* gsomix going to classes about computer networks. | 08:46 | |
gsomix | cu | 08:46 |
gsomix | *is going | 08:46 |
sonne|work | n4nd0: which timezone are you in? | 09:08 |
sonne|work | gsomix: which timezone are you in? | 09:08 |
sonne|work | blackburn: timezone? | 09:08 |
sonne|work | I need to collect to know when you all are online... | 09:08 |
blackburn | sonne|work: both me and gsomix are in +4 | 09:09 |
n4nd0 | sonne|work: CEST (UTC+2) | 09:09 |
n4nd0 | sonne|work: I think we are in the same | 09:10 |
sonne|work | n4nd0: yes indeed | 09:10 |
sonne|work | does anyone know which timezone pluskid is in? | 09:10 |
blackburn | +8 | 09:10 |
sonne|work | and heiko? | 09:10 |
blackburn | I guess +1? | 09:11 |
n4nd0 | btw, if someone had any doubt, apart from being pretty fast, JL's cover tree gives correct results :) | 09:11 |
sonne|work | heh | 09:11 |
sonne|work | nice | 09:11 |
sonne|work | n4nd0: so then it is even more clear that we want it :) | 09:11 |
n4nd0 | sonne|work: I'm working on that ;) | 09:12 |
sonne|work | blackburn: wiking? | 09:12 |
sonne|work | and jacob? | 09:13 |
blackburn | sonne|work: belgium, same? | 09:13 |
blackburn | no idea about jacob | 09:13 |
sonne|work | harshit? and philippe? | 09:13 |
blackburn | harshit +5 or +6 | 09:13 |
n4nd0 | sonne|work: philippe is in France so I guess he has the same | 09:13 |
blackburn | philippe france | 09:13 |
sonne|work | lets better make sure when they pop up | 09:15 |
-!- av3ngr [av3ngr@nat/redhat/x-ynyjtavjttwlbxfx] has quit [Ping timeout: 264 seconds] | 09:40 | |
-!- wiking [~wiking@78-23-191-201.access.telenet.be] has joined #shogun | 09:42 | |
-!- wiking [~wiking@78-23-191-201.access.telenet.be] has quit [Changing host] | 09:43 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 09:43 | |
sonne|work | wiking: time zone? | 09:44 |
wiking | cest | 09:44 |
sonne|work | that is UTC+2 right? | 09:44 |
sonne|work | 9:45hrs now correct? | 09:45 |
wiking | y | 09:45 |
wiking | yes | 09:45 |
sonne|work | kthx | 09:46 |
blackburn | seems you guys are having shortest word competition | 09:47 |
wiking | :> | 09:47 |
sonne|work | I've updated the website http://shogun-toolbox.org/ | 10:18 |
sonne|work | not sure if a github banner is a good idea but hey I added it for now | 10:19 |
sonne|work | also removed broken links to tuebingen mpg | 10:19 |
blackburn | yes it is | 10:19 |
n4nd0 | nice | 10:20 |
sonne|work | ahh but I got the link wrong | 10:20 |
blackburn | sonne|work: we need new description I think | 10:20 |
sonne|work | hah | 10:20 |
sonne|work | after gsoc :) | 10:20 |
sonne|work | and volunteers welcome! | 10:20 |
n4nd0 | what description are you talking about guys? | 10:21 |
blackburn | n4nd0: shogun is blablabla | 10:21 |
n4nd0 | the paragraph on top of the page? | 10:21 |
blackburn | yeah | 10:21 |
n4nd0 | ok | 10:21 |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking] | 10:43 | |
-!- pluskid [~chatzilla@li400-235.members.linode.com] has joined #shogun | 10:48 | |
n4nd0 | what about int and float in JL's code, should they go to int32_t and float64_t, respectively? | 10:48 |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 10:49 | |
blackburn | sonne|work: it seems jacob is in eastern usa - msu.edu stands for michigan state univ | 10:50 |
sonne|work | no overlap in this set I guess then | 10:51 |
sonne|work | not good | 10:51 |
blackburn | sonne|work: may be some saturday evening would work | 10:51 |
blackburn | however it is saturday heh | 10:52 |
pluskid | cool to hear good news about 9 slots! | 10:52 |
sonne|work | pluskid: you were UTC+8 right? | 10:52 |
pluskid | GMT+8, is it the same as UTC+8? | 10:53 |
pluskid | I'll google it | 10:53 |
blackburn | yes | 10:53 |
pluskid | sonne|work: yes, UTC+8 | 10:53 |
pluskid | the homepage github link point to http://github.com/you , is it intended? | 10:56 |
blackburn | lol | 10:57 |
pluskid | and the "you" actually has 16 followers | 10:57 |
pluskid | though no public repo | 10:57 |
pluskid | Previously I thought one mentor can be on multiple project, am I wrong? | 10:59 |
blackburn | you mean one person? | 11:00 |
pluskid | yes | 11:00 |
blackburn | yes that can be | 11:00 |
pluskid | one person (as a mentor) can be in several projects(ideas), isn't it? | 11:01 |
pluskid | I'm not sure | 11:01 |
sonne|work | pluskid: yes | 11:01 |
sonne|work | but we try to restrict things | 11:01 |
sonne|work | to be 1vs1 | 11:01 |
pluskid | oh, that might be good for better quality | 11:02 |
sonne|work | which means 7-8 slots is what we will actually use | 11:03 |
sonne|work | (the downside of things) | 11:03 |
pluskid | :-/ | 11:03 |
wiking | blackburn: hey man i'm testing the multi class linear machine... and i'm not soooo satisfied with the speed :D | 11:07 |
blackburn | wiking: which one? | 11:08 |
wiking | classification is super fast :) | 11:08 |
wiking | MulticlassLibLinear | 11:08 |
blackburn | how much time? d? N? | 11:08 |
blackburn | if you do modelselection enable save train state option | 11:09 |
blackburn | to make training a little faster | 11:09 |
-!- pluskid [~chatzilla@li400-235.members.linode.com] has quit [Quit: ChatZilla 0.9.88.2 [Firefox 11.0/20120312181643]] | 11:10 | |
blackburn | wiking: what is epsilon? | 11:13 |
wiking | 1e-5 | 11:13 |
blackburn | uh that is tough | 11:13 |
blackburn | try larger no need to 1e-5 | 11:13 |
wiking | 1e-2? | 11:14 |
blackburn | yeah 1e-2 or 1e-3 | 11:14 |
wiking | it's funny | 11:16 |
wiking | i have a new kernel | 11:16 |
wiking | but it only works with libsvm | 11:16 |
wiking | other svm die because it cannot find enough support vectors :( | 11:16 |
wiking | but with libsvm it works nice | 11:16 |
-!- pluskid [~chatzilla@111.120.11.250] has joined #shogun | 11:17 | |
blackburn | wiking: is it proper normalized then? | 11:18 |
wiking | blackburn: you mean the kernel? yes | 11:19 |
blackburn | strange | 11:19 |
wiking | sqrtdiagkernelnormalizer.. | 11:19 |
wiking | any suggestion for metal earners? | 11:21 |
wiking | hahahah metalearners :) | 11:21 |
wiking | stupid autocorrection | 11:21 |
wiking | :> | 11:21 |
blackburn | metal earner lol | 11:21 |
blackburn | wiking: what kind of metalearner? | 11:22 |
wiking | well anything that does a consensus in a good way between various ML classifications | 11:22 |
wiking | because i have some good results with different features.... separately | 11:22 |
wiking | but when i add them together (even with combined kernel) the result degrades :) | 11:23 |
blackburn | uh | 11:23 |
wiking | so it's pretty annoying :) | 11:24 |
sonne|work | wiking: then your normalization sucks | 11:24 |
wiking | sonne|work: hahahahaha any suggestions ? :) | 11:25 |
pluskid | blackburn: I find why LinearRidgeRegression is acting weird :) | 11:25 |
blackburn | pluskid: why? | 11:25 |
pluskid | a small bug | 11:25 |
pluskid | the matrices are not initialized | 11:25 |
pluskid | simple fix :p | 11:25 |
sonne|work | :) | 11:26 |
blackburn | sonne|work: bad boy | 11:26 |
sonne|work | wiking: well do you have the data or do I have it ... | 11:26 |
* sonne|work smacks sonney2k | 11:26 | |
wiking | :>>> | 11:26 |
* sonne|work luckily I am at work | 11:26 | |
pluskid | :> | 11:26 |
pluskid | btw I have some suggestions about CMath::twonorm and several related functions | 11:30 |
pluskid | the template function is returning the same type as the argument | 11:30 |
pluskid | for example, when you calculate twonorm for an int32_t vector | 11:30 |
pluskid | you get int32_t result, which would loss a lot of information | 11:31 |
wiking | pluskid: yeah | 11:31 |
wiking | ah ok | 11:31 |
pluskid | I suggest we use float64_t or something for all input types | 11:31 |
wiking | i think in case of qnorm is not like that | 11:31 |
wiking | afaik | 11:31 |
pluskid | why? | 11:31 |
pluskid | have to leave for a while | 11:32 |
wiking | mmm wait i'm mixing now | 11:32 |
pluskid | cu later | 11:32 |
wiking | ah yeah | 11:32 |
wiking | qnorm is like that as well | 11:32 |
wiking | shouldn't be it should be something like pluskid says | 11:32 |
wiking | there's anyways an explicit casting there | 11:34 |
wiking | when u return float64_t as class T | 11:34 |
sonne|work | wiking: in case when we compute a float64_t anyways yes | 11:35 |
wiking | man it's early here in my head | 11:35 |
wiking | i meant implicit casting.. | 11:35 |
wiking | but anyhow yeah it should always return something like float64_t even if the input vector is an int32_t | 11:35 |
sonne|work | wiking: there is always a trade-off | 11:41 |
sonne|work | this can be wanted or not | 11:41 |
sonne|work | when numbers are small the pure int32_t based compuation will use MMX and be very fast | 11:41 |
sonne|work | and correct | 11:41 |
sonne|work | if numbers become too big one needs the float64_t based variant | 11:42 |
sonne|work | computation is slow but this is required for large numbers | 11:42 |
sonne|work | so YMMV | 11:42 |
pluskid | in our current implementation, float64_t is always used in computaion | 11:46 |
pluskid | just when we are returnining, it gets casted to T | 11:46 |
pluskid | I think the trade-off sonne|work mentioned is true | 11:46 |
pluskid | but what we are discussing here is not that issue | 11:46 |
pluskid | we are thinking about whether we should return T or float64_t ? | 11:47 |
-!- gsomix [~gsomix@188.168.2.126] has quit [Ping timeout: 276 seconds] | 12:07 | |
n4nd0 | have a GSoC applicant just arrived? | 12:12 |
n4nd0 | a new one I mean | 12:12 |
wiking | ? | 12:13 |
blackburn | n4nd0: what? | 12:14 |
n4nd0 | mailing list | 12:14 |
n4nd0 | brb | 12:15 |
blackburn | ahh+ | 12:15 |
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has quit [Quit: Page closed] | 12:22 | |
-!- pluskid [~chatzilla@111.120.11.250] has quit [Ping timeout: 246 seconds] | 12:36 | |
-!- pluskid [~chatzilla@173.254.214.60] has joined #shogun | 12:37 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: Lost terminal] | 13:18 | |
-!- vikram360 [~vikram360@117.192.171.231] has quit [Ping timeout: 252 seconds] | 13:26 | |
-!- vikram360 [~vikram360@117.192.164.248] has joined #shogun | 13:27 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 13:28 | |
-!- vikram360 [~vikram360@117.192.164.248] has quit [Ping timeout: 272 seconds] | 13:36 | |
-!- vikram360 [~vikram360@117.192.160.228] has joined #shogun | 13:37 | |
-!- gsomix [~gsomix@188.168.4.7] has joined #shogun | 13:42 | |
-!- gsomix [~gsomix@188.168.4.7] has quit [Read error: Connection reset by peer] | 13:43 | |
-!- genix [~gsomix@188.168.4.7] has joined #shogun | 13:43 | |
genix | hello | 13:45 |
-!- genix is now known as gsomix | 13:45 | |
gsomix | It's hot. (~o___o)~ | 13:47 |
gsomix | n4nd0, what about covertree? | 13:48 |
-!- PhilTillet [~Philippe@157.159.42.154] has joined #shogun | 13:48 | |
n4nd0 | gsomix: on progress | 13:48 |
gsomix | sonne|work, what about slots? | 13:48 |
sonne|work | gsomix: what about SG_PROGRESS :P | 13:48 |
PhilTillet | hey everybody | 13:49 |
n4nd0 | haha | 13:49 |
sonne|work | PhilTillet: what is your timezone? | 13:50 |
PhilTillet | sonne|work, France, why? :p | 13:51 |
sonne|work | that's part of the decision process | 13:52 |
PhilTillet | oki ^^ | 13:52 |
sonne|work | so? | 13:53 |
PhilTillet | Hmm, I think it is called CET | 13:53 |
PhilTillet | or GMT+2 | 13:54 |
sonne|work | so it is 13:54 ? | 13:54 |
PhilTillet | yes | 13:54 |
sonne|work | thx | 13:55 |
PhilTillet | you're welcome :) | 13:57 |
-!- PhilTillet [~Philippe@157.159.42.154] has quit [Ping timeout: 246 seconds] | 14:29 | |
wiking | what's a good kernel for sparse features apart from linear? | 14:37 |
n4nd0 | well guys I have the first timing results of JL cover tree used from shogun! | 14:41 |
n4nd0 | someone expecting :P? | 14:42 |
pluskid | show the data :D | 14:43 |
n4nd0 | http://snipt.org/uhOi1#expand | 14:44 |
n4nd0 | there is still some space for improvement | 14:44 |
n4nd0 | so it is promising | 14:44 |
pluskid | wow | 14:44 |
pluskid | tree construction plus querying | 14:44 |
n4nd0 | yeah, everything there | 14:44 |
pluskid | significantly outperforms qsort | 14:44 |
pluskid | correctness of the results validated? | 14:45 |
n4nd0 | not for this case I have timed | 14:45 |
pluskid | anyway, it's super cool | 14:46 |
n4nd0 | I have validated them for the non-integrated version, they were the same | 14:46 |
pluskid | cool | 14:46 |
n4nd0 | so I guess that they must remain, it is just the class used for the cover tree points what changes | 14:46 |
pluskid | I think I should suggest some guys in my lab to try out cover tree in their projects :D | 14:46 |
n4nd0 | :D | 14:46 |
n4nd0 | I have still some work left to do | 14:47 |
n4nd0 | I wanted to see some results fast | 14:47 |
pluskid | sure | 14:47 |
pluskid | encouraging! | 14:47 |
n4nd0 | pluskid: I wanted to ask you by the way | 14:47 |
n4nd0 | for you accuracy evaluation for KNN | 14:47 |
pluskid | what do you mean? | 14:48 |
n4nd0 | I think I could use it to check that everything is alright using the cover tree, right? | 14:48 |
n4nd0 | let me see, maybe I am not using the right words | 14:48 |
pluskid | I can't quite catch your meaning | 14:48 |
n4nd0 | I don't really remember the exact words | 14:48 |
n4nd0 | the first patch you did for shogun | 14:48 |
pluskid | accuracy evaluation? | 14:49 |
n4nd0 | yeah | 14:49 |
n4nd0 | then it is what I said ;) | 14:49 |
pluskid | yes | 14:49 |
n4nd0 | is that something that can be used to check if the results given by KNN are correct? | 14:49 |
pluskid | why not simply comparing the two results? | 14:50 |
pluskid | covertree and qsort | 14:50 |
n4nd0 | yeah sure | 14:50 |
n4nd0 | that's the way I do it | 14:50 |
n4nd0 | I just wondered if your patch was something "more profesional" to do this | 14:50 |
pluskid | if they are the same, the algorithm should be correct | 14:50 |
pluskid | I don't think so | 14:50 |
n4nd0 | ok | 14:50 |
pluskid | my patch is for clustering | 14:50 |
pluskid | here is classification | 14:50 |
n4nd0 | ok, my bad :S | 14:50 |
pluskid | I think if two algorithms are the same, they should give the same accuracy | 14:51 |
pluskid | but when they give the same accuracy, they are not necessarily the same :p | 14:51 |
sonne|work | n4nd0: buy you can of course use any of the other evaluation modules that where made for clustering :D | 14:51 |
n4nd0 | sonne|work: what do you think about the result? | 14:51 |
sonne|work | very lovely :) | 14:52 |
n4nd0 | sonne|work: but we can make it better | 14:52 |
sonne|work | that is factor 4 now right? | 14:52 |
n4nd0 | sonne|work: yeah | 14:52 |
sonne|work | hmmhh wasn't knn even using multiple threads? | 14:52 |
n4nd0 | I have not configured anything for threads | 14:53 |
n4nd0 | so I think it is just with one | 14:53 |
n4nd0 | the cover tree admits an heuristic to compute the distance | 14:54 |
n4nd0 | there is an upper bound | 14:54 |
n4nd0 | while computing the distance, if the intermediate result is already higher than this upper bound, one can cut the comptuations | 14:54 |
n4nd0 | computations, sorry | 14:54 |
n4nd0 | right now I am not doing that, I am just using the method distance in CDistance | 14:55 |
n4nd0 | sonne|work: what about including in CDistance this distance computation with the upper bound? | 14:55 |
sonne|work | n4nd0: feel free... | 14:55 |
n4nd0 | sonne|work: cool | 14:55 |
n4nd0 | I have still some time to have fun with this :D | 14:56 |
n4nd0 | I will keep you update | 14:56 |
sonne|work | it is actually not so easy | 14:56 |
n4nd0 | sonne|work: why? | 14:56 |
sonne|work | you would need to modify *every* distance function | 14:56 |
n4nd0 | my idea was | 14:56 |
n4nd0 | just to define it in Euclidian | 14:56 |
n4nd0 | do people use KNN with other than that? | 14:56 |
sonne|work | to stop doing computations once a certain distance is reached | 14:57 |
sonne|work | probably | 14:57 |
sonne|work | but agreed euclidian is most important | 14:57 |
n4nd0 | I can include the heuristic in Euclidian for the moment | 14:58 |
n4nd0 | for the others it would work just fine | 14:58 |
-!- PhilTillet [~Philippe@npasserelle10.minet.net] has joined #shogun | 15:01 | |
-!- vikram360 [~vikram360@117.192.160.228] has quit [Ping timeout: 245 seconds] | 15:14 | |
-!- harshit_ [~harshit@182.68.45.192] has joined #shogun | 15:19 | |
n4nd0 | see you later people | 15:29 |
sonne|work | cu | 15:30 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 15:31 | |
pluskid | cu | 15:32 |
gsomix | sonne|work, i have just returned from uni. | 15:38 |
gsomix | but... done | 15:38 |
gsomix | sonne|work, so what about slots? | 15:38 |
pluskid | browse irc log ;) | 15:39 |
sonne|work | gsomix: ahh 9 but we will return 1 | 15:41 |
sonne|work | so 8 | 15:41 |
sonne|work | which is still *a lot* | 15:41 |
gsomix | sonne|work, is it the final decision? or should google make sure that shogun could have 8 students? | 15:45 |
sonne|work | ? | 15:45 |
sonne|work | I don't understand | 15:45 |
harshit_ | sonne|work: Do you remember the wrapper code for SVMlin.cpp. | 15:46 |
sonne|work | harshit_: hardly | 15:46 |
sonne|work | btw which timezone are you in? | 15:46 |
harshit_ | Actually I am making its TSVM available in shogun,I have dont most of the part , but optimization is really slow | 15:46 |
harshit_ | done* | 15:47 |
gsomix | sonne|work, awww, I'm stupid. Never mind, I already figured out. | 15:48 |
sonne|work | ?!!? | 15:49 |
sonne|work | harshit_: timezone? | 15:49 |
-!- blackburn [~qdrgsm@83.234.54.186] has joined #shogun | 15:51 | |
CIA-64 | shogun: Jacob Walker master * rb98500e / src/shogun/regression/GaussianProcessRegression.cpp : Hopefully fixed strange memory bug in CGaussianProcessRegression::GetCovarianceMatrix - http://git.io/RLy04g | 15:52 |
CIA-64 | shogun: Sergey Lisitsyn master * rb95b089 / src/shogun/regression/GaussianProcessRegression.cpp : Merge branch 'master' of git://github.com/puffin444/shogun - http://git.io/1cKr-g | 15:52 |
harshit_ | about +5:30 | 15:55 |
harshit_ | GMT | 15:55 |
harshit_ | btw congrats on getting your requested no of slots, perhaps even more :) | 15:55 |
sonne|work | harshit_: thx | 15:56 |
-!- pluskid [~chatzilla@173.254.214.60] has quit [Quit: ChatZilla 0.9.88.2 [Firefox 11.0/20120314124128]] | 15:56 | |
sonne|work | does any one of you blog? | 15:58 |
-!- pluskid [~pluskid@173.254.214.60] has joined #shogun | 15:59 | |
-!- pluskid [~pluskid@173.254.214.60] has quit [Client Quit] | 16:00 | |
gsomix | cu, guys | 16:01 |
blackburn | I do not, only twitter | 16:01 |
-!- pluskid [~pluskid@173.254.214.60] has joined #shogun | 16:01 | |
gsomix | sonne|work, too only twitter | 16:02 |
harshit_ | sonne|work: not yet. | 16:03 |
sonne|work | he | 16:04 |
wiking | blackburn: what's your twitter acc | 16:04 |
blackburn | wiking: @qdrgsm | 16:05 |
wiking | sonne|work: i do but it's really unrelated to what i do :P | 16:05 |
shogun-buildbot | build #644 of r_static is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/r_static/builds/644 | 16:05 |
wiking | blackburn: will be funny to read your twits :) | 16:05 |
blackburn | oh I twit some shit :D | 16:06 |
wiking | yeah i can see that | 16:06 |
wiking | :>>>> | 16:06 |
sonne|work | hurray | 16:07 |
wiking | build success? :)) | 16:07 |
-!- romi_ [~mizobe@187.66.121.180] has quit [Remote host closed the connection] | 16:07 | |
sonne|work | yeah | 16:07 |
wiking | who uses shogun from r? :) | 16:09 |
wiking | i mean i dislike the whole R as it is so ... ;) | 16:09 |
shogun-buildbot | build #666 of cmdline_static is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cmdline_static/builds/666 | 16:09 |
wiking | ok | 16:10 |
wiking | any ideas who could i debug | 16:10 |
wiking | why i don't get support vectors for certain svm-s? | 16:10 |
wiking | with my kernel? | 16:10 |
wiking | gnmpsvm for that matter... | 16:11 |
gsomix | >> build #666 >:] | 16:11 |
-!- pluskid [~pluskid@173.254.214.60] has quit [Quit: leaving] | 16:12 | |
-!- pluskid [~pluskid@173.254.214.60] has joined #shogun | 16:12 | |
sonne|work | wiking: likely trivial solution is better... | 16:14 |
shogun-buildbot | build #645 of octave_static is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/octave_static/builds/645 | 16:14 |
harshit_ | Good news , TSVM is working,just need some modifications and examples and then people can use unlabeled data as well ,!!! | 16:15 |
wiking | sonne|work: ? | 16:15 |
sonne|work | can happen for noise data that just outputting a constant is better | 16:15 |
wiking | mmm | 16:16 |
wiking | but libsvm works fine that's the strange thing | 16:16 |
wiking | with same features and same kernel | 16:16 |
wiking | everything is the same just the optimizer is different | 16:17 |
shogun-buildbot | build #632 of python_static is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/python_static/builds/632 | 16:19 |
sonne|work | wait one is a true multiclass svm the other not | 16:20 |
sonne|work | so it is a big difference | 16:20 |
harshit_ | blackburn: How can we print SG_SDEBUG() and SG_SINFO() messages in octave_modular | 16:20 |
blackburn | harshit_: anyobject.io.set_loglevel(MSG_DEBUG) | 16:22 |
harshit_ | thanks :) | 16:22 |
-!- harshit_ [~harshit@182.68.45.192] has quit [Quit: Leaving] | 16:36 | |
CIA-64 | shogun: Soeren Sonnenburg master * r2d7f35d / src/configure : | 16:38 |
CIA-64 | shogun: use ${SWIG} instead of explicit swig in configure | 16:38 |
CIA-64 | shogun: thanks Paul B?nau for the report - http://git.io/3IoNUA | 16:38 |
-!- blackburn [~qdrgsm@83.234.54.186] has quit [Quit: Leaving.] | 16:51 | |
-!- blackburn [5bde8018@gateway/web/freenode/ip.91.222.128.24] has joined #shogun | 17:04 | |
@sonney2k | wiking, you still havent' documented things http://shogun-toolbox.org/buildbot/builders/python_modular/builds/468/steps/compile/logs/warnings%20%2826%29 | 17:07 |
@sonney2k | ? | 17:07 |
wiking | hehehe i have a paper deadline tonight at 2am :))) | 17:08 |
@sonney2k | aha | 17:08 |
@sonney2k | but time to hang around here I see. must be important ;-) | 17:08 |
-!- nickon [~noneedtok@dD5774105.access.telenet.be] has joined #shogun | 17:09 | |
wiking | buuut | 17:10 |
@sonney2k | gsomix, what's up with the SG_PROGRESS patch? | 17:10 |
wiking | well i cannot make faster the cpus | 17:10 |
wiking | so in the meanwhile it's calculating :) | 17:10 |
-!- vikram360 [~vikram360@117.192.160.228] has joined #shogun | 17:10 | |
wiking | heheh yeah actually it's a cvpr paper should be important :>>. | 17:10 |
@sonney2k | ahh and PhilTillet, you did the nearest centroid right - so there are also some warnings in there http://shogun-toolbox.org/buildbot/builders/python_modular/builds/468/steps/compile/logs/warnings%20%2826%29 | 17:11 |
@sonney2k | for you | 17:11 |
@sonney2k | wiking, I see - shouldn't then you *write* the paper? | 17:11 |
wiking | heheheh it's done | 17:11 |
wiking | need to fill it out with stats | 17:11 |
@sonney2k | heh | 17:12 |
wiking | pull request sent | 17:14 |
wiking | ;) | 17:14 |
PhilTillet | sonney2k, true, i'll fix that asap | 17:15 |
blackburn | sonney2k is in 'work, slaves!' mood today | 17:17 |
wiking | :)))) | 17:18 |
CIA-64 | shogun: Viktor Gal master * r9af8f41 / src/shogun/preprocessor/PNorm.h : Fix doxygen comment for getter/setter functions - http://git.io/C9Bu2g | 17:19 |
CIA-64 | shogun: Soeren Sonnenburg master * r87b031d / src/shogun/preprocessor/PNorm.h : | 17:19 |
CIA-64 | shogun: Merge pull request #443 from vigsterkr/master | 17:19 |
CIA-64 | shogun: Fix doxygen comment for PNorm - http://git.io/-9YHFQ | 17:19 |
@sonney2k | thx | 17:19 |
@sonney2k | gtg... | 17:19 |
-!- blackburn [5bde8018@gateway/web/freenode/ip.91.222.128.24] has quit [Ping timeout: 245 seconds] | 17:47 | |
-!- pluskid [~pluskid@173.254.214.60] has quit [Read error: Connection reset by peer] | 18:02 | |
gsomix | sonney2k, I'm corrected. before I asked about the slots. :3 | 18:42 |
-!- blackburn1 [~qdrgsm@83.234.54.186] has joined #shogun | 18:59 | |
wiking | thanks god for deadline extensions ! \o/ | 19:01 |
@sonney2k | more time to waste on experiments | 19:02 |
@sonney2k | why do they do this? | 19:02 |
@sonney2k | gsomix, ? | 19:02 |
@sonney2k | blackburn1, what does gsomix mean to say? | 19:02 |
wiking | sonney2k: ;D | 19:02 |
blackburn1 | lol | 19:03 |
blackburn1 | let me check logs | 19:03 |
@sonney2k | blackburn1, regarding slavery - long abandoned :-D but all these things I keep repeating to do take like 5 minutes of work and I mentioned them a couple of times already - so I want them to be done. | 19:03 |
blackburn1 | sonney2k: oh you take these words seriously ;) | 19:04 |
-!- blackburn1 is now known as lisitsyn | 19:04 | |
lisitsyn | ok let it be so | 19:04 |
lisitsyn | :D | 19:04 |
lisitsyn | who stole my nickname | 19:05 |
lisitsyn | ah long stayed web connection | 19:06 |
lisitsyn | sonney2k: ok he said he is corrected however no idea what does it mean :D | 19:07 |
lisitsyn | nevermind he just surprised with 9 slots | 19:07 |
@sonney2k | arent we all | 19:08 |
@sonney2k | it is like x-mas, easter, birthday all on one day | 19:09 |
lisitsyn | haha | 19:09 |
gsomix | wut? | 19:09 |
lisitsyn | shogun is a serious thing for you ;) | 19:09 |
@sonney2k | now I only hope 3 students stick with us after this gsoc | 19:10 |
gsomix | sonney2k, awww. my English is very bad. I mean, that I corrected SG_PROGRESS patch. | 19:10 |
@sonney2k | gsomix, ahhh! | 19:10 |
lisitsyn | ohh | 19:10 |
lisitsyn | even I didn't get it | 19:10 |
gsomix | I use too many passive voice. hehe :( | 19:11 |
@sonney2k | so that we finally have some kind of developer community | 19:11 |
lisitsyn | sonney2k: have you seen page generator at github? | 19:11 |
lisitsyn | I can't use it while I am only pushermaan | 19:11 |
gsomix | lisitsyn, where is blackburn? | 19:12 |
lisitsyn | nickname is in use by some hanged webchat connection | 19:12 |
-!- lisitsyn is now known as shogun-blackburn | 19:12 | |
shogun-blackburn | do you like it more? | 19:12 |
wiking | sonney2k: the fork me label on the page was cool! | 19:12 |
-!- shogun-blackburn is now known as blackburn_ | 19:13 | |
gsomix | shogun-blackburn, it sounds like bot. | 19:13 |
CIA-64 | shogun: Soeren Sonnenburg master * r19f81a5 / src/shogun/classifier/GaussianNaiveBayes.cpp : | 19:13 |
CIA-64 | shogun: Merge pull request #439 from gsomix/sgprogress | 19:13 |
CIA-64 | shogun: add SG_PROGRESS messages to Gaussian Naive Bayes - http://git.io/ZeFwTA | 19:13 |
blackburn_ | gsomix: LIKE A BOSS | 19:14 |
gsomix | blackburn_, huh. | 19:14 |
@sonney2k | blackburn_, you remain vodking | 19:14 |
blackburn_ | sonney2k: why should I ever stop | 19:14 |
blackburn_ | http://www.youtube.com/watch?feature=player_detailpage&v=NisCkxU544c | 19:15 |
blackburn_ | commit changes! like a boss! promote shogun! like a boss! github workflow! like a boss! | 19:15 |
blackburn_ | my own classes! like a boss! | 19:15 |
gsomix | I heard word `vodka'. | 19:16 |
blackburn_ | gsomix: I want you to write an essay 'how my life changed when I start shoguning and vodking' | 19:18 |
gsomix | blackburn_, okay, master. =___= | 19:18 |
@sonney2k | blackburn_, hilarious | 19:18 |
PhilTillet | lol the chan is going crazy :p | 19:25 |
@sonney2k | I think the only sane participant is shogun-buildbot | 19:26 |
PhilTillet | maybe one day the buildbots will have enough artificial intelligence to get crazy aswell :D | 19:27 |
@sonney2k | lets hope they don't listen | 19:29 |
@sonney2k | blackburn_, can we merge pluskid's LARS etc patch? | 19:30 |
blackburn_ | sonney2k: I didn't take a closer look yet but as it is already tested - probably we can | 19:33 |
blackburn_ | sonney2k: how would you like to solve Subset problem? | 19:35 |
blackburn_ | no typemap for index_t | 19:35 |
blackburn_ | so should be a typemap or it should be int32_t as it is but without index_t | 19:35 |
blackburn_ | ah you commented it already | 19:36 |
CIA-64 | shogun: Sergey Lisitsyn master * rba71ded / (2 files): Exposed Subset in modular #445 - http://git.io/DMzgTQ | 19:38 |
-!- harshit_ [~harshit@182.68.181.141] has joined #shogun | 19:42 | |
-!- blackburn_ [~qdrgsm@83.234.54.186] has quit [Quit: Leaving.] | 19:48 | |
-!- blackburn1 [~qdrgsm@83.234.54.186] has joined #shogun | 19:52 | |
@sonney2k | index_t is int32_t so it should just work | 20:07 |
blackburn1 | sonney2k: yes but [1,2,3] is probably not int32_t | 20:14 |
@sonney2k | that is why I cast things | 20:15 |
blackburn1 | sonney2k: have you seen 'refactoring' PR? | 20:16 |
@sonney2k | yeah | 20:17 |
@sonney2k | not sure what to do with it | 20:17 |
blackburn1 | sonney2k: I have no words to comment it :D | 20:17 |
blackburn1 | sonney2k: there are some PRs sometimes you can stare and stare and stare and nothing to say | 20:18 |
@sonney2k | parts are ok parts are not - and it is soo huge that I don't know | 20:18 |
@sonney2k | for example I like the const declarations instead of #defines | 20:18 |
@sonney2k | but then it is too big to be merged at once | 20:19 |
blackburn1 | sonney2k: are you ok some code uses stl vector in LARS? | 20:28 |
blackburn1 | I am not against it while it can be easily refactored next day | 20:28 |
@sonney2k | blackburn1, yes - just tell him | 20:34 |
-!- mohataher [c4cd9077@gateway/web/freenode/ip.196.205.144.119] has joined #shogun | 20:34 | |
blackburn1 | ok | 20:34 |
mohataher | hi, I applied for a GSoC, anywyas, I was working on implementing Convex Multi-class Feature Learning algorithm and I didn't really find any alternative functions in C++ equivalent to Diag(k) in matlab | 20:36 |
mohataher | any suggestions ? | 20:36 |
mohataher | if not, I can manage to implement and test it otherwise | 20:36 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 20:40 | |
n4nd0 | hi all | 20:40 |
blackburn1 | mohataher: we have one in CMath afaik however it is not any complex | 20:41 |
blackburn1 | n4nd0: hey | 20:41 |
mohataher | blackburn1 : I have just looked up the class, seems like I didn't find any functions equivalent to Diag(k), do you know the name of the function in Math.h ? | 20:51 |
blackburn1 | mohataher: create_diagonal_matrix | 20:52 |
mohataher | Is it added recently ? my version doesn't have this function, I will update my version and get back to you if I needed. Thanks | 20:57 |
blackburn1 | mohataher: depends how old is your version ;) | 21:00 |
-!- mohataher [c4cd9077@gateway/web/freenode/ip.196.205.144.119] has quit [Ping timeout: 245 seconds] | 21:10 | |
gsomix | btw, congratulations on the day of space exploration. | 21:25 |
gsomix | http://www.youtube.com/watch?v=FG0fTKAqZ5g | 21:34 |
shogun-buildbot | build #207 of nightly_none is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_none/builds/207 | 21:56 |
@sonney2k | blackburn1, have you heard anything from jacob? | 21:58 |
gsomix | shogun-buildbot, good work! | 21:58 |
shogun-buildbot | What you say! | 21:58 |
blackburn1 | sonney2k: anything like? | 21:59 |
@sonney2k | you merged his patch but was he here in the channel? | 21:59 |
* sonney2k is trying to figure out the timezone he is in | 21:59 | |
blackburn1 | sonney2k: yes | 21:59 |
blackburn1 | sonney2k: michigan | 21:59 |
@sonney2k | blackburn1, what is his nick? | 22:00 |
blackburn1 | hmm | 22:02 |
blackburn1 | sonney2k: just puffin444 | 22:02 |
blackburn1 | http://www.shogun-toolbox.org/irclogs/%23shogun.2012-04-11.log.html | 22:02 |
blackburn1 | sonney2k: check 18-48 | 22:02 |
@sonney2k | so yesterday | 22:07 |
@sonney2k | I was hoping he is here today | 22:07 |
@sonney2k | blackburn1, btw I tried to return a slot but got an internal server error :D | 22:07 |
blackburn1 | haha | 22:07 |
blackburn1 | sonney2k: but he fixed the error | 22:07 |
blackburn1 | however I hope it is not his last commit heh | 22:08 |
@sonney2k | oliver said that what he did was 10-20% of the gsoc work | 22:09 |
@sonney2k | but yes agreed | 22:09 |
@sonney2k | :) | 22:09 |
-!- harshit_ [~harshit@182.68.181.141] has quit [Ping timeout: 240 seconds] | 22:09 | |
blackburn1 | sonney2k: 10-20%? seems to be small project | 22:12 |
@sonney2k | like all of our projects ;-) | 22:12 |
@sonney2k | cu | 22:13 |
@sonney2k | ZZZZZZzzzzzzoo.... | 22:13 |
n4nd0 | sonney2k: good night | 22:13 |
blackburn1 | sonney2k: no I am planning more that I can :( | 22:14 |
-!- nickon [~noneedtok@dD5774105.access.telenet.be] has quit [Quit: ( www.nnscript.com :: NoNameScript 4.22 :: www.esnation.com )] | 22:15 | |
-!- harshit_ [~harshit@182.68.181.141] has joined #shogun | 22:22 | |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking] | 22:25 | |
-!- wiking [~wiking@78-23-191-201.access.telenet.be] has joined #shogun | 22:37 | |
-!- wiking [~wiking@78-23-191-201.access.telenet.be] has quit [Changing host] | 22:37 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 22:37 | |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Remote host closed the connection] | 22:37 | |
-!- wiking [~wiking@vpnb162.ugent.be] has joined #shogun | 22:37 | |
-!- wiking [~wiking@vpnb162.ugent.be] has quit [Changing host] | 22:38 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 22:38 | |
wiking | blackburn1: here? | 22:44 |
blackburn1 | da | 22:44 |
wiking | cool | 22:44 |
wiking | so whoever made gmm is around? | 22:45 |
wiking | Alesis Novik | 22:45 |
blackburn1 | no, he is doing phd in england and seems to be pretty busy | 22:45 |
wiking | i was wondering if anybody would mind if it would be possible that u could preset the kmeans clusters | 22:45 |
wiking | when you want to do gmm | 22:45 |
blackburn1 | kind of warm start? | 22:46 |
wiking | yep | 22:46 |
wiking | e.g. u already have a preknowledge about the clusters. | 22:46 |
blackburn1 | I don't mind if you do that ;) | 22:46 |
wiking | ok | 22:47 |
wiking | since now i'm testing gmm | 22:47 |
wiking | on a very small dataset | 22:47 |
wiking | and it's horribly slow | 22:47 |
wiking | and as a matter of fact i do have the cluster centers from kmeans... | 22:48 |
wiking | so i thought maybe it'd be good to save some time.. | 22:48 |
wiking | but first i'll add some debugging code to gmm | 22:48 |
wiking | and dtrace it or something | 22:49 |
wiking | to see where goes the most calc | 22:49 |
wiking | nyiha | 22:52 |
wiking | most of the cpu time is used by dlasr | 22:52 |
wiking | which is basically max_likelihood calc | 22:56 |
blackburn1 | dlasr?? | 22:58 |
wiking | it's from lapack | 22:58 |
blackburn1 | yes I know that function | 23:00 |
blackburn1 | I mean I am surprised ;) | 23:00 |
-!- harshit_ [~harshit@182.68.181.141] has quit [Quit: Leaving] | 23:18 | |
gsomix | good night, guys | 23:21 |
n4nd0 | good night gsomix | 23:23 |
n4nd0 | nice video the one of the space by the way ;) | 23:23 |
-!- blackburn1 [~qdrgsm@83.234.54.186] has quit [Ping timeout: 244 seconds] | 23:33 | |
-!- blackburn [~qdrgsm@83.234.54.186] has joined #shogun | 23:35 | |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Ping timeout: 252 seconds] | 23:54 | |
--- Log closed Fri Apr 13 00:00:19 2012 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!