--- Log opened Tue Mar 06 00:00:19 2012 | ||
@sonney2k | n4nd0, IIRC the regularization term just adds sth to the diagonal of the cov matrix | 00:06 |
---|---|---|
@sonney2k | so it penalizes non-zero weights in w | 00:08 |
@sonney2k | n4nd0, I would say have a look at QDA in scikits learn and follow the reference above | 00:09 |
@sonney2k | n4nd0, btw it would make a lot of sense to create one graphical example illustrating lda/qda/ lda on polyfeatures | 00:09 |
@sonney2k | so we see how they differ :) | 00:10 |
@sonney2k | at least I would love to see that | 00:10 |
@sonney2k | so yes QDA is of interest especially when just running LDA on PolyFeatures | 00:11 |
@sonney2k | for comparison | 00:12 |
n4nd0 | sonney2k: I have been studying the code from scikits before, I think I got ir more or less | 00:14 |
n4nd0 | I have not seen the direct relation to the theory I read though | 00:15 |
n4nd0 | they use SVD to compute the covariances what I found surprising at first sight | 00:16 |
n4nd0 | would you like to have sth graphical like this? | 00:16 |
n4nd0 | http://scikit-learn.org/0.5/auto_examples/plot_lda_vs_qda.html | 00:16 |
n4nd0 | http://scikit-learn.org/stable/auto_examples/plot_lda_qda.html | 00:16 |
-!- blackburn [~qdrgsm@188.168.5.46] has joined #shogun | 00:17 | |
@sonney2k | n4nd0, yes sth like this | 00:18 |
n4nd0 | sonney2k: I will work on that then :) | 00:19 |
@sonney2k | n4nd0, do they use SVD for computing the inverse? | 00:19 |
@sonney2k | pseudo inverse I mean | 00:20 |
n4nd0 | sonney2k: mmm I am not sure, by the name of their variables it looks like they compute the covariance, not the inverse | 00:20 |
blackburn | eigs of XX' are left singular vectors | 00:21 |
blackburn | eigvectors I mean | 00:22 |
blackburn | damn are you crazy guys? pretty late | 00:22 |
n4nd0 | it is this fragment of code where they do it http://snipt.org/ufHf0 | 00:22 |
n4nd0 | blackburn: haha my gf was sick today so I have been wandering around home almost the whole day and didn't work that much :P | 00:23 |
blackburn | oh let she recover soon | 00:23 |
@sonney2k | n4nd0, we compute pseudo inverse for LDA too via SVD | 00:24 |
@sonney2k | /The pseudo inverse A+ can be constructed from the singular value | 00:24 |
@sonney2k | /decomposition A = UDV^T , by A^+ = V(D+)U^T. | 00:24 |
@sonney2k | (fromt he code) | 00:24 |
@sonney2k | what blackburn said | 00:24 |
n4nd0 | sonney2k: ok | 00:25 |
blackburn | I had some article on pseudoinverse | 00:25 |
blackburn | sonney2k: what have I said? | 00:25 |
@sonney2k | blackburn, seen that http://www.h-online.com/open/news/item/GitHub-security-incident-highlights-Ruby-on-Rails-problem-1463207.html ? | 00:25 |
n4nd0 | sonney2k: in any case I think that the whole point of computing the covariance and its inverse in QDA is to compute the Mahalanobis distance | 00:25 |
n4nd0 | sonney2k: so I thought that maybe we could just use MahalanobisDistance for that | 00:26 |
blackburn | sonney2k: hah crazy | 00:26 |
@sonney2k | blackburn, on #gsoc I had a chat with carol smith (gsoc organizer) and indeed the tutorial task seems to be out of scope for gsoc | 00:29 |
blackburn | sonney2k: damn.. | 00:30 |
@sonney2k | I asked her if we can rescue this part somehow, i.e. by modifying it to only have code parts in there like: write/polish examples, enable test suite based on these examples, fix code to have fewer ASSERT's etc | 00:31 |
@sonney2k | lets see if that is possible | 00:31 |
n4nd0 | ups ... not good | 00:32 |
-!- blackburn [~qdrgsm@188.168.5.46] has quit [Ping timeout: 265 seconds] | 00:47 | |
-!- blackburn [d55780cc@gateway/web/freenode/ip.213.87.128.204] has joined #shogun | 00:48 | |
blackburn | sonney2k: still there? | 00:48 |
blackburn | n4nd0: still there? | 00:51 |
n4nd0 | blackburn: yes | 00:51 |
n4nd0 | blackburn: I drank coffee :P | 00:51 |
blackburn | too much? | 00:51 |
blackburn | ;) | 00:51 |
n4nd0 | blackburn: haha just a small cup, I am don't drink it usually so ... | 00:52 |
blackburn | n4nd0: if you find qda too messy we may come up with some other idea? | 00:53 |
n4nd0 | n4nd0: I think it can be ok, I must learn to do new stuff so it is ok even if is a bit difficult to get it at the beginning | 00:55 |
n4nd0 | blackburn: or do you say because you would like it to take it yourself? | 00:55 |
blackburn | n4nd0: no, I don't want to for sure :) | 00:55 |
blackburn | I'm working on mc ocas this week probably ;) | 00:55 |
n4nd0 | blackburn: ok, then I will continue with it | 00:56 |
n4nd0 | blackburn: I have a couple of exams next week so I might work a bit slow but I won't leave it | 00:56 |
blackburn | I think I can suggest you to implement some dim reduction algorithm | 00:56 |
blackburn | no problem just keep in touch | 00:57 |
blackburn | so you won't have any exams in may or later? | 00:57 |
n4nd0 | blackburn: sure, tell me about that dim reduction if you have thought of it already | 00:57 |
n4nd0 | blackburn: yes I will :S | 00:57 |
blackburn | may be t-SNE | 00:57 |
n4nd0 | I have to take a couple of exams from Spain in late April early | 00:58 |
n4nd0 | May | 00:58 |
blackburn | I will have some in may as well | 00:58 |
blackburn | and bachelor's work defence (?) | 00:58 |
n4nd0 | I did a bit of mess with my studies | 00:58 |
n4nd0 | I came last year in Erasmus to KTH, in my 4th year of university | 00:58 |
n4nd0 | in Spain the studies are normally 5 years long so I had still one year left after my Erasmus | 00:59 |
blackburn | I see | 00:59 |
n4nd0 | but I got the opportunity of staying for a master at KTH ... and I didn't want to refuse it | 00:59 |
n4nd0 | so here I am, doing the master and taking courses from my degree in Spain | 00:59 |
blackburn | so you are a msc student? | 01:00 |
n4nd0 | yes | 01:00 |
n4nd0 | but I don't have a real bachelor | 01:00 |
blackburn | I did not get even bachelor yet ;) | 01:00 |
n4nd0 | in Spain it is 5 years so it is like if we get bachelor + msc when we graduate | 01:00 |
blackburn | so you are probably older than me :) | 01:00 |
n4nd0 | I say it is like because from my point of view it is not like that | 01:00 |
n4nd0 | I am 22 | 01:01 |
blackburn | yeah, I'm 20 currently ;) | 01:01 |
n4nd0 | I am from November 1989 | 01:01 |
n4nd0 | you go pretty well actually | 01:01 |
blackburn | btw we have had similar system for a while | 01:02 |
n4nd0 | :) | 01:02 |
blackburn | about 5 year | 01:02 |
n4nd0 | I don't like ... it sucks | 01:02 |
blackburn | but now we have bachelors too | 01:02 |
blackburn | and I'm going to be one of | 01:02 |
n4nd0 | it if 5 years with really broad studies ... we don't really learn to do useful stuff of anything | 01:02 |
n4nd0 | we learn a bit of everything and master nothing :( | 01:03 |
blackburn | is it better in swe? | 01:03 |
n4nd0 | I like it more yes | 01:03 |
n4nd0 | from my point of view in Spain the hard part is to pass the exams | 01:04 |
n4nd0 | but the labs are really easy normally | 01:04 |
n4nd0 | here the exams are quite ok, they test that you get the point of the course, basic exercises | 01:04 |
n4nd0 | but the labs require lot of work | 01:04 |
blackburn | I have two algos to suggest to implement to you | 01:05 |
n4nd0 | tell me | 01:05 |
blackburn | maximum variance unfolding and t-SNE | 01:05 |
blackburn | first one is simple | 01:05 |
n4nd0 | do you have a good reference in a paper or a book for them? | 01:06 |
blackburn | n4nd0: in fact we would even create some idea for dimensionality reduction.. | 01:06 |
blackburn | matlab code is provided in the matlab dimensionality reduction toolbox | 01:06 |
blackburn | and some paper may be | 01:07 |
blackburn | hmm I need to talk with soeren :) | 01:07 |
blackburn | I guess it can be an idea and I could help smb with mentoring | 01:08 |
n4nd0 | ok | 01:08 |
blackburn | n4nd0: broad choice for you :) | 01:08 |
-!- cronor [~cronor@e178177151.adsl.alicedsl.de] has joined #shogun | 01:09 | |
blackburn | uhuhuh | 01:09 |
blackburn | 4.09 already and I have to be on my job on 10.00 damnation! | 01:10 |
n4nd0 | haha | 01:10 |
n4nd0 | I think you should get some sleep | 01:10 |
blackburn | agree | 01:10 |
blackburn | cronor: sorry is it your pull req? | 01:11 |
blackburn | if so - we need some time for this issue :) | 01:11 |
cronor | blackburn: huh? | 01:11 |
blackburn | cronor: then not your :) | 01:12 |
cronor | blackburn: probably yes | 01:12 |
blackburn | n4nd0: okay see you | 01:12 |
n4nd0 | blackburn: good night | 01:12 |
blackburn | good night | 01:13 |
-!- blackburn [d55780cc@gateway/web/freenode/ip.213.87.128.204] has quit [Quit: Page closed] | 01:13 | |
-!- cronor [~cronor@e178177151.adsl.alicedsl.de] has left #shogun [] | 01:44 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 02:43 | |
-!- vikram360 [~vikram360@117.192.190.106] has quit [Read error: Connection reset by peer] | 02:53 | |
-!- vikram360 [~vikram360@117.192.190.106] has joined #shogun | 02:54 | |
-!- vikram360 [~vikram360@117.192.190.106] has quit [Ping timeout: 276 seconds] | 03:17 | |
-!- Zoe [a66f43f2@gateway/web/freenode/ip.166.111.67.242] has joined #shogun | 04:10 | |
-!- Zoe [a66f43f2@gateway/web/freenode/ip.166.111.67.242] has left #shogun [] | 04:12 | |
-!- cronor [~cronor@e178177151.adsl.alicedsl.de] has joined #shogun | 05:16 | |
-!- vikram360 [~vikram360@117.192.160.61] has joined #shogun | 05:27 | |
-!- vikram360 [~vikram360@117.192.160.61] has quit [Ping timeout: 272 seconds] | 05:33 | |
-!- vikram360 [~vikram360@117.192.181.1] has joined #shogun | 05:33 | |
-!- vikram360 [~vikram360@117.192.181.1] has quit [Ping timeout: 244 seconds] | 05:57 | |
-!- vikram360 [~vikram360@117.192.185.120] has joined #shogun | 05:57 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 06:11 | |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking] | 06:32 | |
-!- cronor [~cronor@e178177151.adsl.alicedsl.de] has quit [Quit: cronor] | 08:01 | |
-!- vikram360 [~vikram360@117.192.185.120] has quit [Ping timeout: 260 seconds] | 08:02 | |
-!- vikram360 [~vikram360@117.192.176.91] has joined #shogun | 08:02 | |
-!- vikram360 [~vikram360@117.192.176.91] has quit [Ping timeout: 252 seconds] | 08:33 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 08:48 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 09:12 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Ping timeout: 265 seconds] | 09:54 | |
-!- _dr [~johannes@faui32o.informatik.uni-erlangen.de] has joined #shogun | 10:06 | |
-!- n4nd0 [~nando@n189-p80.kthopen.kth.se] has joined #shogun | 10:33 | |
CIA-64 | shogun: Soeren Sonnenburg master * rcdd0218 / src/shogun/distance/MahalanobisDistance.h : fix latex error in documentation (x_i not \x_i) - http://git.io/PM2A6Q | 11:28 |
n4nd0 | sonne|work: ups sorry for that typo :( | 11:28 |
n4nd0 | sonne|work: one question about QDA, where does it make more sense in the class hierarchy? | 11:29 |
sonne|work | n4nd0: no that was my typo! | 11:30 |
sonne|work | no need to be sorry for the mistakes I make | 11:30 |
n4nd0 | sonne|work: ah ok! :) | 11:30 |
n4nd0 | sonne|work: I think it should go directly under CMachine | 11:30 |
n4nd0 | sonne|work: I have not seen other quadratic classifiers in the toolbox | 11:31 |
sonne|work | indeed ... | 11:32 |
n4nd0 | so it is ok doing CQDA directly under CMachine or something in between like CQuadraticMachine analogous to CLinearMachine? | 11:33 |
sonne|work | just do it directly for now | 11:38 |
sonne|work | as long as we don't have any other quadratic machine it doesn't make sense to be overly general | 11:39 |
n4nd0 | I see, ok then :) | 11:40 |
-!- Netsplit *.net <-> *.split quits: naywhayare | 13:20 | |
-!- Netsplit over, joins: naywhayare | 13:23 | |
-!- vikram360 [~vikram360@117.192.190.93] has joined #shogun | 13:36 | |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Read error: Connection reset by peer] | 15:23 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 15:23 | |
-!- n4nd0 [~nando@n189-p80.kthopen.kth.se] has quit [Ping timeout: 252 seconds] | 15:23 | |
-!- n4nd0 [~nando@n189-p80.kthopen.kth.se] has joined #shogun | 15:41 | |
-!- n4nd0 [~nando@n189-p80.kthopen.kth.se] has quit [Ping timeout: 252 seconds] | 16:19 | |
-!- muddo [~muddo@gateway/tor-sasl/muddo] has quit [Ping timeout: 276 seconds] | 17:39 | |
-!- muddo [~muddo@gateway/tor-sasl/muddo] has joined #shogun | 17:45 | |
-!- puneetgoyal [~puneet@115.242.12.226] has joined #shogun | 18:41 | |
vikram360 | What is the Software Engineering approach that Shogun developers usually tend to take when building a new feature into shogun? Is a prototype first built, tested and then integrated | 18:43 |
vikram360 | Is there any strict methodology that all developers have to follow when building a feature | 18:44 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 18:44 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Ping timeout: 252 seconds] | 18:50 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 18:51 | |
-!- blackburn [~qdrgsm@188.168.2.36] has joined #shogun | 18:52 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Ping timeout: 244 seconds] | 18:56 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 19:08 | |
-!- blackburn [~qdrgsm@188.168.2.36] has quit [Ping timeout: 255 seconds] | 19:16 | |
-!- blackburn [~qdrgsm@83.234.54.251] has joined #shogun | 19:28 | |
blackburn | sonney2k: around? | 19:40 |
-!- in3xes [~in3xes@180.149.49.230] has joined #shogun | 20:20 | |
-!- in3xes [~in3xes@180.149.49.230] has quit [Ping timeout: 260 seconds] | 20:25 | |
-!- in3xes [~in3xes@180.149.49.230] has joined #shogun | 20:25 | |
-!- in3xes [~in3xes@180.149.49.230] has quit [Ping timeout: 252 seconds] | 20:33 | |
-!- in3xes [~in3xes@180.149.49.230] has joined #shogun | 20:49 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Read error: Operation timed out] | 20:50 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 20:52 | |
-!- blackburn [~qdrgsm@83.234.54.251] has quit [Ping timeout: 245 seconds] | 21:31 | |
-!- in3xes [~in3xes@180.149.49.230] has quit [Ping timeout: 240 seconds] | 21:34 | |
-!- puneetgoyal [~puneet@115.242.12.226] has quit [Quit: Leaving] | 22:12 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Ping timeout: 255 seconds] | 22:27 | |
@sonney2k | vikram360, really depends - you could write an external class + some main method and just link to libshogun from c++ side and then when you feel comfortable submit your extension to core shogun providing some examples and interface integration - since we use these examples also as regression tests everything will be fine then. | 22:49 |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking] | 23:33 | |
--- Log closed Wed Mar 07 00:00:19 2012 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!