--- Log opened Wed Apr 24 00:00:53 2019 00:46 < fenn> http://blog.samaltman.com/the-merge 01:15 < LooCfur> fenn, that could be interesting, but what is the accuracy by which that model identifies criminals/non-criminals? 01:21 -!- adamsky [~adamsky3@178235181086.unknown.vectranet.pl] has joined ##hplusroadmap 01:46 < fenn> 90% accuracy 01:49 < LooCfur> fenn, with trained data, or separate data? 01:50 < LooCfur> also, can the way the picture was taken be what you're training on? Like maybe the height of the camera 01:51 < LooCfur> I could see identifying a criminal by how they act and move. But look, separate from style or how the picture was taken? I'm skeptical 01:51 < fenn> i don't know how they tested it. and no, they're not detecting light-dependent features. if you'd bothered to look at the paper you'd know that 01:51 < LooCfur> touche 01:52 < fenn> they're measuring distance between eyes, curvature of upper lip, and some other face measurement 01:52 < LooCfur> the upper lip thing is very distinct in the example pictures, lol. 01:53 < LooCfur> that might be the expression of a normal person going to jail though :P 01:54 < LooCfur> fenn, how does eye distance play in? My eyes are too close together. :P 01:54 < fenn> please try not to make sarcastic comments on politically charged discussions 01:57 < fenn> they are ID photos, not mug shots 02:06 < LooCfur> Yeah, it's kind of interesting. I notice the four sub types for criminal faces are faces I associate with being less happy people. The non-criminals are more "happy". Something of interest: Try it on falsely identified criminals that were later proven innocent. Is it capturing the criminality or the lifestyle? 02:09 < LooCfur> I am surprised that eye distance plays a role. I think I even recall that it didn't, in the past. Although, I do associate a closer distance with being a predator. 02:09 < fenn> ok maybe i'm misunderstanding what the 90% refers to 02:09 < fenn> it's a "true positive rate" 02:12 -!- Gurkenglas [~Gurkengla@unaffiliated/gurkenglas] has joined ##hplusroadmap 02:13 < fenn> meaning, i think, if you test a criminal the identifier will flag them correctly 90% of the time. however, since only 0.36% of chinese are criminals, if you pick a random person off the street there is a 7% probability the classifier will test possible. meaning, if the classifier flags you, a random person, there is only a 4.4% chance you will actually commit a crime (and be convicted) 02:13 < fenn> s/possible/positive/ 02:15 < LooCfur> nice, 4.4% chance you will actually be convicted of committing a crime. (Regardless of whether you did it) though, right? 02:17 -!- jtimon [~quassel@181.61.134.37.dynamic.jazztel.es] has joined ##hplusroadmap 02:17 < fenn> they want to get more ID photos of convicted criminals but can't because of the media shitstorm around the first paper 02:17 < fenn> i don't know much about how stereotypes and biases factor into chinese legal proceedings, but i doubt it's perfectly fair 02:17 < LooCfur> try ID photos before and ID photos after being incarcerated too 02:18 < fenn> these facial features wouldn't change 02:18 < fenn> and it wouldn't be interesting anyway 02:18 < LooCfur> the curve of the upper lip can, I think 02:18 < LooCfur> eye distance is the only thing I don't see changing 02:18 < fenn> the other metric was the ratio of tip of the nose to width of mouth 02:19 < fenn> er, more specifically the angle 02:20 < LooCfur> "criminals" have more of a pouty mouth. I would guess that lowers mouth width 02:20 < fenn> i doubt being in prison would simultaneously make one's nose both pointier and less pointy 02:20 < fenn> the point here is that criminal faces show more variance 02:25 < juul> have you all seen the crowdfunding campaign for open source software for the HiSeq 2000/2500 ? 02:26 < juul> https://wemakeit.com/projects/reseq-reuse-dna-sequencers?locale=en 02:35 -!- Urchin [~urchin@unaffiliated/urchin] has joined ##hplusroadmap 02:37 < fenn> nice microscope :) 02:54 -!- cannedprimates [sid16585@gateway/web/irccloud.com/x-cllcuzaetsdwsvwc] has quit [Ping timeout: 264 seconds] 02:54 -!- cannedprimates_ [sid16585@gateway/web/irccloud.com/x-gowcgkapmoyiudiv] has joined ##hplusroadmap 03:20 < juul> yeah i'd love to have access to a microscope like that 03:27 -!- adamsky [~adamsky3@178235181086.unknown.vectranet.pl] has quit [Remote host closed the connection] 03:28 -!- adamsky [~adamsky3@178235181086.unknown.vectranet.pl] has joined ##hplusroadmap 03:37 -!- Urchin [~urchin@unaffiliated/urchin] has quit [Ping timeout: 258 seconds] 03:43 -!- yoleaux [~yoleaux@xn--ht-1ia18f.nonceword.org] has quit [Ping timeout: 258 seconds] 04:44 -!- Urchin [~urchin@unaffiliated/urchin] has joined ##hplusroadmap 04:53 -!- jtimon [~quassel@181.61.134.37.dynamic.jazztel.es] has quit [Quit: gone] 04:59 -!- LooCfur [~hello@loocfur.powered.by.lunarbnc.net] has quit [Read error: Connection reset by peer] 05:00 -!- LooCfur [~hello@loocfur.powered.by.lunarbnc.net] has joined ##hplusroadmap 05:39 -!- yoleaux [~yoleaux@xn--ht-1ia18f.nonceword.org] has joined ##hplusroadmap 07:40 -!- spaceangel [~spaceange@ip-86-49-16-65.net.upcbroadband.cz] has joined ##hplusroadmap 07:45 -!- Gurkenglas [~Gurkengla@unaffiliated/gurkenglas] has quit [Read error: Connection reset by peer] 07:52 -!- adamsky [~adamsky3@178235181086.unknown.vectranet.pl] has quit [Ping timeout: 250 seconds] 07:59 -!- strages [uid11297@gateway/web/irccloud.com/x-qpytofvazelhppig] has joined ##hplusroadmap 09:21 -!- sachy [~sachy@78.108.102.220] has joined ##hplusroadmap 11:00 -!- Urchin [~urchin@unaffiliated/urchin] has quit [Ping timeout: 268 seconds] 12:22 < nmz787> "This is to be done in Java using the Micro-Manager plugin framework." -- welp, I'm out 12:23 < fenn> it's the de-facto standard microscope control software 12:31 < nmz787> huh, never heard of it 12:31 < nmz787> microscopists generally like imageJ though, I know 12:32 < nmz787> (and I know that's Java) 13:56 -!- LeoTal [~Adium@91.247.7.93.rev.sfr.net] has quit [Ping timeout: 258 seconds] 14:07 -!- LooCfur [~hello@loocfur.powered.by.lunarbnc.net] has quit [Quit: Free ZNC ~ Powered by LunarBNC: https://LunarBNC.net] 14:13 -!- LooCfur [~hello@loocfur.powered.by.lunarbnc.net] has joined ##hplusroadmap 14:17 -!- spaceangel [~spaceange@ip-86-49-16-65.net.upcbroadband.cz] has quit [Remote host closed the connection] 14:34 < kanzure> https://wemakeit.com/projects/reseq-reuse-dna-sequencers "We want to write free and open source software so we people re-use the HiSeq 2000 and 2500 DNA sequencers and have a modest crowd-funder going." 14:34 < kanzure> oh damn it juul 15:23 -!- delinquentme [~delinquen@12.88.222.194] has joined ##hplusroadmap 15:35 -!- LooCfur [~hello@loocfur.powered.by.lunarbnc.net] has quit [Quit: Free ZNC ~ Powered by LunarBNC: https://LunarBNC.net] 15:36 -!- delinquentme [~delinquen@12.88.222.194] has quit [Quit: Leaving] 15:47 -!- adamsky [~adamsky3@178235181086.unknown.vectranet.pl] has joined ##hplusroadmap 15:49 -!- bbrittain is now known as bwb_ 15:50 -!- adamsky [~adamsky3@178235181086.unknown.vectranet.pl] has quit [Remote host closed the connection] 16:37 < kanzure> "Multifunctional CRISPR-Cas9 with engineered immunosilenced human T cell epitopes" https://www.nature.com/articles/s41467-019-09693-x 17:13 -!- l_wl [~l_wl@pool-108-45-141-95.washdc.fios.verizon.net] has quit [Remote host closed the connection] 17:14 -!- l_wl [~l_wl@pool-108-45-141-95.washdc.fios.verizon.net] has joined ##hplusroadmap 17:15 < periffis> kanzure: has justanotheruser ever asked you dumb shit like "if we became a cloned-but-gene-edited species, would you think we'd keep JUST MEN or JUST WOMEN around" 17:40 -!- Gurkenglas [~Gurkengla@unaffiliated/gurkenglas] has joined ##hplusroadmap 17:43 -!- mrdata-- [~mrdata@unaffiliated/mrdata] has joined ##hplusroadmap 18:26 -!- strages [uid11297@gateway/web/irccloud.com/x-qpytofvazelhppig] has quit [Quit: Connection closed for inactivity] 18:46 -!- Gurkenglas [~Gurkengla@unaffiliated/gurkenglas] has quit [Ping timeout: 250 seconds] 18:58 -!- indiebio [~quassel@2a05:d018:c1e:8d00:c473:4128:ae2d:2c36] has quit [Quit: http://quassel-irc.org - Chat comfortably. Anywhere.] 18:58 -!- indiebio [~quassel@2a05:d018:c1e:8d00:c473:4128:ae2d:2c36] has joined ##hplusroadmap 19:00 -!- mrdata-- [~mrdata@unaffiliated/mrdata] has quit [Read error: Connection reset by peer] 19:05 -!- periffis [~g@66.205.193.233] has quit [Quit: periffis] 19:49 -!- l_wl [~l_wl@pool-108-45-141-95.washdc.fios.verizon.net] has quit [] 19:49 -!- TC [~talinck@98.29.27.253] has joined ##hplusroadmap 19:50 -!- TC is now known as Guest98078 19:53 -!- hehelleshin [~talinck@98.29.27.253] has quit [Ping timeout: 258 seconds] 20:05 -!- LeoTal [~Adium@121.247.7.93.rev.sfr.net] has joined ##hplusroadmap 20:52 -!- ebowden [~ebowden@unaffiliated/ebowden] has quit [Remote host closed the connection] 21:20 < kanzure> periffis: no 21:30 -!- periffis [~g@66.205.193.233] has joined ##hplusroadmap 21:30 -!- periffis [~g@66.205.193.233] has quit [Read error: Connection reset by peer] 21:31 -!- periffis [~g@66.205.193.233] has joined ##hplusroadmap 22:04 -!- LooCfur [~hello@loocfur.powered.by.lunarbnc.net] has joined ##hplusroadmap 22:06 -!- mark_otaris [firrre@wikimedia/Mark-Otaris] has quit [Ping timeout: 264 seconds] 22:06 -!- mark_otaris [~mark_otar@wikimedia/Mark-Otaris] has joined ##hplusroadmap 22:23 -!- BitMEX_Chad [~BitMEX_Ch@101.188.169.213] has joined ##hplusroadmap 23:00 -!- BitMEX_Chad [~BitMEX_Ch@101.188.169.213] has quit [] 23:58 -!- Livestradamus [~Livestrad@unaffiliated/livestradamus] has quit [Quit: Livestradamus] --- Log closed Thu Apr 25 00:00:44 2019