Open in new window / Try shogun cloud
--- Log opened Fri Jan 06 00:00:19 2012
CIA-1shogun: Sergey Lisitsyn master * ra8869b0 / (13 files): Updated doc of dimensionality reduction toolkit -
CIA-1shogun: Sergey Lisitsyn master * r2e2788c / (2 files): Fixed typos -
-!- blackburn [~qdrgsm@] has left #shogun []05:00
-!- blackburn [~qdrgsm@] has joined #shogun14:24
blackburnsonne|work: around?14:57
blackburnsonney2k: sonne|work any of you? ;)21:01
@sonney2kblackburn, wassup?21:02
blackburnsonney2k: check dis out
blackburnshould be distributed alongside with my paper :)21:03
blackburnjust after I finish it..21:03
@sonney2klot of effort21:04
blackburnI told you I want to write a 'book', I guess I started it accidentally21:04
@sonney2kgo ahead and do that for the rest of shogun ;-)21:04
blackburnsure, I like this kind of work21:04
blackburnwell I love latex21:04
blackburntoo much I guess21:04
blackburnsonney2k: where do you think I should include this doc?21:06
@sonney2kblackburn, well we could create shogun/tutorial21:07
blackburnhmm yes sure21:07
@sonney2kand then below that have shogun/tutorial/dimensionreduction21:07
blackburnI'm worried what to include along with paper21:07
blackburnthe whole shogun?..21:08
@sonney2kI don't understand21:08
blackburnsonney2k: jmlr mloss track?21:08
blackburnI need to include software21:08
blackburnmy paper..21:08
blackburnabout dim reduction stuff21:08
@sonney2kwhat is the problem?21:09
blackburnwhat should I include?21:09
blackburnwhole shogun?21:09
@sonney2kI would say yes21:09
blackburnthat's pretty strange that's why I'm worried with it21:10
@sonney2kwhat are the alternatives?21:10
@sonney2kcreate a big patch file and say - please apply to shogun 1.1.0 ?21:10
blackburnno alternatives21:11
blackburnsonney2k: can we release 1.1.1 when I finish including this paper stuff?21:12
@sonney2kI would like to have heikos migrations in the next release21:13
blackburni think it will take 1-2 weeks more21:13
blackburnwhat is heiko's plan?21:14
@sonney2kbtw, I wanted to propose a different way of getting code / patches merged21:14
@sonney2klike you / me also only submitting pull requests21:14
@sonney2kand me/you reviewing them21:14
blackburnsonney2k: yeah you told it before.. but I still think it will only take our time21:15
@sonney2kit will improve quality - we both make mistakes the other one can spot easily21:58
@sonney2kso it would help - however we should have some hard deadlines until when a patch has to be reviews21:59
@sonney2ke.g. <24hrs21:59
@sonney2kotherwise one is free to merge21:59
@sonney2kblackburn, ^]21:59
blackburnhmm we should try..22:00
blackburnsonney2k: could you please take a look on paper 4/5 of my paper? are graphs ok?22:01
blackburnpaper of paper22:01
@sonney2kblackburn, yes figures look great22:03
blackburnsonney2k: should I enlarge text?22:04
blackburnoh it already takes too much22:04
@sonney2kmaybe a bit - but not essential for initial version22:04
blackburnsonney2k: ok thanks22:04
@sonney2kit is good that you renamed it to efficient dim red toolbox22:05
@sonney2klarge scale would make many people upset22:05
@sonney2kI know it is fast but it requires the distance matrix to be in memory for most algos22:06
@sonney2kthat is a problem22:06
blackburnyeah sure22:07
blackburnwell I renamed it because of you told it22:07
blackburnsonney2k: could you please glance over abstract, I reworked it just now?22:14
blackburnsonney2k: btw I mentioned you in acks ;)22:22
-!- puneetgoyal [~puneetgoy@] has joined #shogun23:45
puneetgoyalblackburn: hi23:45
blackburnpuneetgoyal: hi23:45
puneetgoyalblackburn: I am having words such as '98', 'ss' and some more such words or decimals with high tf-idf values23:46
puneetgoyaland I think that is where the problem is23:46
blackburnyes, it could be a problem23:47
blackburnbut strange '98' has high tf-idf23:48
puneetgoyalyes, I have checked the alog a lot of times23:48
puneetgoyalactually there are some data files for which charset is some different type23:49
puneetgoyallike CHINESEBIG523:49
puneetgoyaland from there such words are coming mainly23:50
blackburnpuneetgoyal: well don't care about it now, just try to classify docs23:52
--- Log closed Sat Jan 07 00:00:19 2012