IRC logs of #shogun for Monday, 2018-06-18

--- Log opened Mon Jun 18 00:00:54 2018
-!- HeikoS [~heiko@2a00:23c5:e10a:5c00:f034:eb87:fc6d:a68c] has joined #shogun00:11
-!- mode/#shogun [+o HeikoS] by ChanServ00:11
-!- HeikoS [~heiko@2a00:23c5:e10a:5c00:f034:eb87:fc6d:a68c] has quit [Quit: Leaving.]00:32
-!- HeikoS [~heiko@2a00:23c5:e10a:5c00:f034:eb87:fc6d:a68c] has joined #shogun00:33
-!- mode/#shogun [+o HeikoS] by ChanServ00:33
-!- HeikoS [~heiko@2a00:23c5:e10a:5c00:f034:eb87:fc6d:a68c] has quit [Ping timeout: 245 seconds]00:57
-!- witness_ [uid10044@gateway/web/irccloud.com/x-tdejqhmbqfeftemc] has joined #shogun03:53
-!- Farouk [9cc294f6@gateway/web/freenode/ip.156.194.148.246] has joined #shogun14:00
Faroukwiking: hi14:00
@wikingFarouk, hello14:01
@wikingok just finished my call14:01
@wikingwe should meet now right?14:01
FaroukGoogle hangouts?14:01
Faroukyea14:01
@wikingyep14:01
@wikingsounds good14:01
@wikingdo you have link?14:01
Faroukalright14:01
Farouklink for ?14:01
@wikinghangout :)14:02
@wikingok i'm gonna get in14:02
@wikingdamn14:02
@wikingg hangout & ff doesn't play nicely14:02
@wikingas usual14:02
@wikinglemme chrome14:02
@sukey[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4335 synchronized by shubham80814:07
@sukey[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4335 synchronized by shubham80814:10
-!- Farouk [9cc294f6@gateway/web/freenode/ip.156.194.148.246] has quit [Ping timeout: 260 seconds]14:50
-!- manish7294 [~yaaic@2405:205:2480:faee:dc1:96a8:2bab:7e18] has joined #shogun15:36
manish7294Hi everyone, I was using knn keeping both training and query dataset same. And it looks like for K = 1, each query point considers it's cooresponding training point as the closest neighbor (having distance as 0 between them, as basically both are the same point) giving accuracy as 100%. Is there a way to avoid this condition so that next nearest neighbor is considered for predicting query point?15:46
manish7294And I assume this will be the same for k = 2, 3,...15:49
@wikingmanish7294, k=1 considers itself15:56
manish7294wiking: Right, can we avoid that?15:57
@wikingthat's the definition :)15:57
manish7294wiking: Understandable, but the problem I am facing is that---this thing is different in mlpack and shogun and I have to generate benchmarks for LMNN, so I was wondering if there's any easy way to do that.15:59
@wikingah15:59
@wikingyou are the one working on lmnn benchmarks :)15:59
manish7294yes15:59
manish7294So can you think of any way, so that both the libraries can be on same page16:00
@wikingmanish7294, i've pinged the right person for your questions in your pr for benchmarking :)16:01
manish7294wiking: I am just coming from there :)16:02
manish7294wiking: Thanks! I thought you were talking of rcurtin16:03
@wikingnono16:03
manish7294:)16:03
@wikingiglesias is the implentor of the lmnnnin shogun16:03
manish7294ya, I typed that message before knowing that you mentioned Iglesias16:04
manish7294sorry for creating confusion :)16:04
@wikingnw16:11
@wikinghe should be your man16:11
manish7294wiking: Right! :) Iglesias have done a pretty good job with LMNN. His work has been very helpful. :)16:16
@wiking:)16:16
wuweiwiking: hi16:29
-!- manish7294 [~yaaic@2405:205:2480:faee:dc1:96a8:2bab:7e18] has left #shogun []16:32
-!- HeikoS [~heiko@bmi-airport-1.inf.ethz.ch] has joined #shogun16:52
-!- mode/#shogun [+o HeikoS] by ChanServ16:52
-!- durovo4 [~durovo@31.8f.559e.ip4.static.sl-reverse.com] has quit [Read error: Connection reset by peer]17:08
-!- durovo [~durovo@31.8f.559e.ip4.static.sl-reverse.com] has joined #shogun17:08
-!- HeikoS [~heiko@bmi-airport-1.inf.ethz.ch] has quit [Quit: Leaving.]19:36
--- Log closed Tue Jun 19 00:00:56 2018

Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!