--- Log opened Tue Aug 08 00:00:10 2017
--- Day changed Tue Aug 08 2017
00:00 -!- augur [~augur@208.66.31.98] has joined ##hplusroadmap
00:03 < archels_> .title http://onlinelibrary.wiley.com/doi/10.1002/adhm.201700214/abstract
00:03 < yoleaux> Novel Cranial Implants of Yttria-Stabilized Zirconia as Acoustic Windows for Ultrasonic Brain Therapy - Gutierrez - 2017 - Advanced Healthcare Materials - Wiley Online Library
00:05 -!- augur [~augur@208.66.31.98] has quit [Ping timeout: 276 seconds]
00:22 -!- esmerelda [~mabel@174-24-228-29.tukw.qwest.net] has quit [Changing host]
00:22 -!- esmerelda [~mabel@unaffiliated/jacco] has joined ##hplusroadmap
00:31 -!- Douhet [~Douhet@unaffiliated/douhet] has quit [Ping timeout: 260 seconds]
00:32 -!- Douhet [~Douhet@unaffiliated/douhet] has joined ##hplusroadmap
00:32 -!- preview_ [~quassel@2407:7000:842d:4000::3] has joined ##hplusroadmap
00:32 -!- preview [~quassel@2407:7000:842d:4000::3] has quit [Read error: No route to host]
00:37 -!- augur [~augur@162.245.20.162] has joined ##hplusroadmap
00:39 -!- CRM114 [~urchin@unaffiliated/urchin] has quit [Ping timeout: 258 seconds]
00:42 -!- augur [~augur@162.245.20.162] has quit [Ping timeout: 240 seconds]
00:53 -!- emeraldgreen [~user@92.222.68.248] has joined ##hplusroadmap
00:53 -!- augur [~augur@162.245.20.162] has joined ##hplusroadmap
00:58 -!- CheckDavid [uid14990@gateway/web/irccloud.com/x-foaxdnomacflmlxw] has joined ##hplusroadmap
00:58 -!- augur [~augur@162.245.20.162] has quit [Ping timeout: 268 seconds]
01:05 -!- poppingtonic1 [~brian@105.63.90.61] has quit [Remote host closed the connection]
01:21 -!- emeraldgreen [~user@92.222.68.248] has quit [Ping timeout: 240 seconds]
01:36 -!- emeraldgreen [~user@92.222.68.248] has joined ##hplusroadmap
01:36 -!- emeraldgreen [~user@92.222.68.248] has quit [Client Quit]
01:36 -!- darsie [~darsie@84-113-55-42.cable.dynamic.surfer.at] has joined ##hplusroadmap
01:41 -!- preview_ is now known as preview
02:01 -!- CRM114 [~urchin@unaffiliated/urchin] has joined ##hplusroadmap
02:05 < maaku> "< ybit> atlanta is looking more and more like the city for me if i'm going to be living in the southern united states" <-- I believe that is the first time I've ever heard that said
02:28 -!- jaboja [~jaboja@jaboja.pl] has joined ##hplusroadmap
02:51 -!- bluebear_ [~dluhos@80.95.97.194] has joined ##hplusroadmap
02:52 -!- preview [~quassel@2407:7000:842d:4000::3] has quit [Remote host closed the connection]
03:10 -!- Burn_ [~Burn@pool-96-241-130-178.washdc.fios.verizon.net] has joined ##hplusroadmap
03:11 -!- Burnin8 [~Burn@pool-96-241-130-178.washdc.fios.verizon.net] has quit [Ping timeout: 268 seconds]
03:12 < kanzure> nmz787: yea that does not seem to be in my files.
03:13 < kanzure> "In this paper, biocompatible materials are proposed for replacing part of the cranium (cranial implants) based on low porosity polycrystalline 8 mol% yttria-stabilized-zirconia (8YSZ) ceramics as acoustic windows for brain therapy. In order to assess the viability of 8YSZ implants to effectively transmit ultrasound, various 8YSZ ceramics with different porosity are tested; their acoustic ...
03:13 < kanzure> ...properties are measured; and the results are validated using finite element models simulating wave propagation to brain tissue through 8YSZ windows. The ultrasound attenuation is found to be linearly dependent on ceramics' porosity. Results for the nearly pore-free case indicate that 8YSZ is highly effective in transmitting ultrasound, with overall maximum transmission efficiency of ≈81%, ...
03:13 < kanzure> ...compared to near total absorption of cranial bone. These results suggest that 8YSZ polycrystals could be suitable acoustic windows for ultrasound brain therapy at 1 MHz."
03:13 < kanzure> uhrmm so the concept is to replace cranial bone with this ceramic and thereby increase ultrasound energy transfer efficiency?
03:17 -!- jrayhawk [~jrayhawk@unaffiliated/jrayhawk] has quit [Ping timeout: 268 seconds]
03:17 -!- jrayhawk [~jrayhawk@unaffiliated/jrayhawk] has joined ##hplusroadmap
03:18 -!- jaboja [~jaboja@jaboja.pl] has quit [Ping timeout: 246 seconds]
03:20 -!- jaboja [~jaboja@jaboja.pl] has joined ##hplusroadmap
03:23 -!- preview [~quassel@2407:7000:842d:4000::3] has joined ##hplusroadmap
03:40 < fltrz> kanzure: theres an infrared (optical) transmission window to the skull, so why don't they use the photoaccoustic effect (flash the light at audio frequencies, absorption of the infrared in the brain by certain chemical species, while the skull is transparent)
03:41 < fltrz> well flash the light at ultrasonic audio frequencies
03:42 < kanzure> because the glorious technique of ultrasound is too compelling http://diyhpl.us/~bryan/papers2/neuro/ultrasound/
03:43 < fltrz> what I describe should produce ultrasound produced within the brain, without invasive cranial windows
03:43 < fltrz> ah but probably not at the intensities described
03:44 < kanzure> re: 'certain chemachenical species' yes there are ultrasound contrast agents that can be delivered by blood that enhance ultrasound energy transfer
03:44 < fltrz> this reads similar: http://diyhpl.us/~bryan/papers2/neuro/ultrasound/Diode%20laser%20generated%20ultrasound%20for%20human%20blood%20cell%20lysis.pdf
03:44 < fltrz> I see
03:45 < fenn> they mean an infrared absorber, not an ultrasound absorber
03:45 < fenn> carbon nanotubes i guess
03:45 < kanzure> also some other things here (not neuroscience related) http://diyhpl.us/~bryan/papers2/ultrasound/
03:46 < fenn> ideally you want a tiny light driven actuator
03:48 < kanzure> in this episode of "weird places that my file server shows up", it seems someone started discussing that orbital ring design paper: https://www.reddit.com/r/space/comments/6qx50a/some_guy_came_up_with_a_concept_for_a_leo/
03:48 < fltrz> "NIR spectrum light takes advantage of the optical window in which skin, tissue, and bone are mostly transparent to NIR light in the spectrum of 700-900 nm, while hemoglobin (Hb) and deoxygenated-hemoglobin (deoxy-Hb) are stronger absorbers of light."
03:48 < fenn> i assume the ceramic cranial implants were to reduce scattering of ultrasound waves by being more isotropic
03:48 < fltrz> so the absorber is already present, ready to be optically stimulated at ultrasound frequencies
03:49 < kanzure> "autotldr" bot trying to summarize a transcript from the server: https://www.reddit.com/r/KIC8462852/comments/501aw9/transcript_of_jason_wrights_892016_talk_at_the/d711za3/
03:50 < kanzure> hmm wtf https://archive.org/details/20131009_diyhpl.us_user_bryan
03:52 < kanzure> my summary of igem projects made it into this presentation? page 35 http://steminnovation.sg/wp-content/uploads/2017/06/Endy_Yale_NUS_STEM_v1.pdf
03:53 < kanzure> and page 40
03:54 < kanzure> oh this is drew endy
03:57 < kanzure> .to nmz787 you will like this one, https://www.google.co.in/search?tbm=pts&hl=en&q=nmz787&gws_rd=ssl
03:57 < yoleaux> kanzure: I'll pass your message to nmz787.
03:57 < kanzure> http://www.crispr.kitchen/microbiome-editing-scenarios/scenario-iii-bacterial-contraception
03:57 < kanzure> http://diyhpl.us/~bryan/papers2/diybio/trojok-bacterial-contraception.pdf
04:04 < kanzure> we show up in this patent: "Recombinant virus and preparations thereof" https://www.google.com/patents/WO2014145599A3?cl=en&hl=en
04:24 -!- CRM114 [~urchin@unaffiliated/urchin] has quit [Ping timeout: 258 seconds]
04:25 -!- Malvolio [~Malvolio@unaffiliated/malvolio] has joined ##hplusroadmap
04:39 -!- berndj [~berndj@mail.azna.co.za] has quit [Ping timeout: 255 seconds]
04:41 -!- berndj [~berndj@mail.azna.co.za] has joined ##hplusroadmap
04:45 -!- CRM114 [~urchin@unaffiliated/urchin] has joined ##hplusroadmap
04:51 -!- augur [~augur@162.245.20.162] has joined ##hplusroadmap
04:56 -!- augur [~augur@162.245.20.162] has quit [Ping timeout: 268 seconds]
05:00 -!- aeiousomething [~aeiousome@183.82.170.54] has joined ##hplusroadmap
05:00 < aeiousomething> hello maaku
05:18 -!- aeiousomething [~aeiousome@183.82.170.54] has quit [Ping timeout: 268 seconds]
06:07 -!- CheckDavid [uid14990@gateway/web/irccloud.com/x-foaxdnomacflmlxw] has quit [Quit: Connection closed for inactivity]
06:08 -!- Storyteller [~Storytell@unaffiliated/storyteller] has joined ##hplusroadmap
06:09 -!- Storyteller [~Storytell@unaffiliated/storyteller] has quit [Max SendQ exceeded]
06:09 -!- Storyteller [~Storytell@unaffiliated/storyteller] has joined ##hplusroadmap
06:10 < Storyteller> good morning
06:11 < Storyteller> no deer in the yard I can see so far, had a doe and fawn yesterday
06:13 -!- aeiousomething [~aeiousome@183.82.170.54] has joined ##hplusroadmap
06:21 -!- adlai [~adlai@unaffiliated/adlai] has quit [Ping timeout: 240 seconds]
06:21 -!- adlai [~adlai@unaffiliated/adlai] has joined ##hplusroadmap
06:23 -!- aeiousomething [~aeiousome@183.82.170.54] has quit [Quit: leaving]
06:26 < Storyteller> this is interesting
06:26 < Storyteller> http://digiconomist.net/bitcoin-energy-consumption
06:27 < kanzure> i think we have already demonstrated that you are incapable of high-quality thinking
06:27 < Storyteller> seems bitcoin uses more power than a small country
06:27 < Storyteller> i think you are demonstrating you are incapable of having a pleasant conversation
06:27 < kanzure> i don't care, get out
06:27 < Storyteller> did you make this channel just to try to feel superior?
06:27 < kanzure> yes
06:27 < Storyteller> are you so weak?
06:28 < kanzure> https://www.youtube.com/watch?v=qk17pcwDAAQ
06:29 < Storyteller> your fear has consumed you. you realize, you are responsible for all that wasted energy. for a currency used to buy weapons and whores for rich princes? you did that, and trying to 'feel superior' will not erase your actions.
06:30 < Storyteller> you enjoy your money son.
06:30 < Storyteller> "Mr Bishop"
06:31 < kanzure> so that's your idea of pleasant? :-)
06:31 < kanzure> you have failed at superior thinking and now you are failing by your own metric.
06:31 -!- mode/##hplusroadmap [+o kanzure] by ChanServ
06:31 < Storyteller> no son, Ive no longer any desire to please you.
06:31 < Storyteller> you arent worth the trouble.
06:32 <@kanzure> then you will leave?
06:32 < Storyteller> i was invited here, and interested in the topic, the only one with a problem, is you. Im sorry you are so challenged by my 'inferior thinking'
06:32 < Storyteller> try diplomacy, instead of despotism
06:33 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has quit [Ping timeout: 248 seconds]
06:33 <@kanzure> i see no reason to practice diplomacy here? you have low quality thinking process that probably infects everything you do.
06:35 < Storyteller> you say that. yet you are wrong. its an easy insult, befitting the likes of you. really, try growing up. you are petulant child throwing a tantrum because I dont like your science fair project. get over it.
06:35 <@kanzure> there are many people that don't like bitcoin. that's fine. you are still an idiot.
06:35 -!- Storyteller was kicked from ##hplusroadmap by kanzure [Storyteller]
06:35 -!- mode/##hplusroadmap [-o kanzure] by kanzure
06:37 < fenn> nick szabo had an essay about anthropolgy of money, and how there were tribes that specialized in making shell beads and basically printed their own money, but it was so time consuming that other tribes used the specialists' beads as their money instead of making their own beads. he goes on to describe how an equilibrium arises such that some percentage of an economy goes towards creating money
06:38 < fenn> tokens
06:38 < fenn> i'm pretty sure the essay was not this one, although it covers most of the same topics http://web.archive.org/web/20160302224312/http://szabo.best.vwh.net/shell.html
06:39 < fenn> i thought bitcoin would consume the planet until i read the elusive essay
06:39 < kanzure> the guy is an idiot, he insisted on replacing a proof-of-work function with "social good" functions (hwich would violate the required properties)
06:40 < kanzure> at some point adversarial malicious intent is indistinguishable from malicious ignorance
06:42 < fenn> yes he is a buzzword-level thinker
06:43 < TMA> fenn: that's an amusing euphemism
06:44 < fenn> well there's not enough data to draw that conclusion just from what he's said in this channel
06:44 < fenn> lots of idiots in #space recently
06:45 < kanzure> it would have been better if he had said something like "i have absolutely no idea what bitcoin is or how it works, but i have this idea for something that uses many of the same words, which i recognize is a stupid way of thinking, but i am going to share it anyway".
06:45 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has joined ##hplusroadmap
06:45 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has quit [Client Quit]
06:46 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has joined ##hplusroadmap
06:47 < kanzure> 15:56 < Storyteller> if you can develop the bit coin, help me develop one
06:47 < kanzure> 15:57 < kanzure> no
07:00 -!- jaboja [~jaboja@jaboja.pl] has quit [Ping timeout: 260 seconds]
07:03 -!- aeiousomething [~aeiousome@183.82.170.54] has joined ##hplusroadmap
07:03 < aeiousomething> test
07:41 < kanzure> "In situ programming of leukaemia-specific T cells using synthetic DNA nanocarriers" http://www.nature.com/nnano/journal/v12/n8/abs/nnano.2017.57.html
08:08 -!- jaboja [~jaboja@jaboja.pl] has joined ##hplusroadmap
08:51 < kanzure> https://www.fastcompany.com/3063187/scifutures-probes-your-companys-dystopian-nightmares-and-dreams-up-solutions  https://news.ycombinator.com/item?id=14958329
08:52 < aeiousomething> kanzure: do you have any data on possible healing of ruptured spinal discs
09:00 < kanzure> there was some liquid metal spinal cord neuron healing stuff, that's all i know
09:04 -!- sachy [~sachy@nat.brmlab.cz] has joined ##hplusroadmap
09:24 < nmz787> anyone here ordered on taobao?
09:24 < yoleaux> 10:57Z <kanzure> nmz787: you will like this one, https://www.google.co.in/search?tbm=pts&hl=en&q=nmz787&gws_rd=ssl
09:26 -!- strages [uid11297@gateway/web/irccloud.com/x-eyymtnnmglqtfrgd] has joined ##hplusroadmap
09:26 < kanzure> er, wrong link
09:26 < kanzure> https://www.google.com/search?tbm=pts&hl=en&q=nmz787&gws_rd=ssl
09:27 < kanzure> fenn: https://scholar.google.com/scholar?q=%22fennetic.net%22&btnG=&hl=en&as_sdt=0%2C44
09:28  * fenn braces for impact
09:30 < fenn> eh not too bad
09:31 < fenn> there's so many stupid patents
09:31 < fenn> "dividing a 3D object into a number of parts and then scheduling these parts separately and/or combining 3D objects, or parts of 3D objects, from various input jobs to be printed at the same time"
09:31 < fenn> wow you really needed to patent that idea!
09:54 -!- danfox [~danfox@vulpinedesigns.co.uk] has quit [Read error: Connection reset by peer]
09:56 -!- danfox [~danfox@vulpinedesigns.co.uk] has joined ##hplusroadmap
09:58 -!- jaboja [~jaboja@jaboja.pl] has quit [Ping timeout: 268 seconds]
10:16 -!- JayDugger [~jwdugger@47.185.237.246] has joined ##hplusroadmap
10:28 -!- jaboja [~jaboja@jaboja.pl] has joined ##hplusroadmap
11:13 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has joined ##hplusroadmap
11:18 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has quit [Ping timeout: 268 seconds]
11:26 -!- jaboja [~jaboja@jaboja.pl] has quit [Ping timeout: 260 seconds]
11:30 -!- jaboja [~jaboja@jaboja.pl] has joined ##hplusroadmap
11:52 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has joined ##hplusroadmap
11:55 -!- darsie [~darsie@84-113-55-42.cable.dynamic.surfer.at] has quit [Quit: . ###physics ##ultimatefreedom ##drugs ##ecology ##ads ##sexology ##nudism ##fringescience .]
11:57  * CRM114 facepalms
12:01 -!- darsie [~darsie@84-113-55-42.cable.dynamic.surfer.at] has joined ##hplusroadmap
12:06 -!- darsie [~darsie@84-113-55-42.cable.dynamic.surfer.at] has quit [Quit: . ###physics ##ultimatefreedom ##drugs ##ecology ##ads ##sexology ##nudism ##fringescience .]
12:19 -!- TC [~talinck@cpe-174-97-113-184.cinci.res.rr.com] has joined ##hplusroadmap
12:19 -!- TC is now known as Guest8591
12:23 -!- hehelleshin [~talinck@cpe-174-97-113-184.cinci.res.rr.com] has quit [Ping timeout: 268 seconds]
12:29 -!- bluebear_ [~dluhos@80.95.97.194] has quit [Quit: Leaving.]
12:34 -!- sachy [~sachy@nat.brmlab.cz] has quit [Quit: Leaving.]
12:36 -!- jaboja [~jaboja@jaboja.pl] has quit [Ping timeout: 240 seconds]
13:11 -!- augur [~augur@noisebridge130.static.monkeybrains.net] has joined ##hplusroadmap
13:29 -!- Malvolio [~Malvolio@unaffiliated/malvolio] has quit [Quit: Malvolio]
13:49 -!- aeiousomething [~aeiousome@183.82.170.54] has quit [Ping timeout: 276 seconds]
14:30 < fltrz> does anyone know of a recent comprehensive review of different AI image synthesis techniques? I'm specifically hoping to find a fast (for generation, not necessarily training) method, even at the expense of other qualities
14:43 -!- Malvolio [~Malvolio@unaffiliated/malvolio] has joined ##hplusroadmap
15:09 < kanzure> "Scribbler: Controlling deep image synthesis with sketch and color" https://arxiv.org/pdf/1612.00835.pdf
15:09 < nmz787> https://www.sciencedaily.com/releases/2017/07/170725154216.htm
15:09 < nmz787> Molecular microscopy illuminates molecular motor motion
15:09 < nmz787> 09:24 < nmz787> anyone here ordered on taobao?
15:09 < nmz787> 09:24 < nmz787> anyone here ordered on taobao?
15:10 < nmz787> "Conquering the Silicon Photonics Production Bottleneck" https://www.photonics.com/Article.aspx?AID=62191
15:10 < nmz787> sorry, paste buffer was being weird
15:10 < kanzure> "Real-time user-guided image colorization with learned deep priors" https://arxiv.org/pdf/1705.02999.pdf
15:11 < kanzure> this one does sketch2anime https://arxiv.org/pdf/1705.01908.pdf
15:12 -!- MoonRabbit [4270498f@gateway/web/freenode/ip.66.112.73.143] has joined ##hplusroadmap
15:14 < kanzure> hand pose estimation https://arxiv.org/pdf/1702.03431.pdf
15:14 < kanzure> more manga machine learning https://arxiv.org/pdf/1706.06918.pdf
15:15 < kanzure> video future frame prediction https://arxiv.org/pdf/1702.04125.pdf
15:18 < kanzure> okay i want something that reads patents and then looks at the images to predict the images (minus all the stupid lines and words)
15:18 < kanzure> and then reads hplusroadmap logs and generates mechanical drawings of our proposals.
15:20 < kanzure> and if i was a total sellout i would do the same for gif generation
15:25 < kanzure> fltrz: for style transfer, ulyanov is like <1 sec per 512x512 image. and dumoulin and belongie.
16:10 < docl> Antarctica based ice gun proposal from Nowicki http://www.nss.org/settlement/nasa/Nowicki/SPBI110.HTM
16:10 < nmz787> so I don't get the whole bolt ring idea... seems pushing a bolt into space would shove the platform back to earth
16:11 < docl> (antarctic ice gun) seems kinda impractical to me, but maybe ice would be a good construction material in a high desert area of the US where you can create snow by spraying water in the air.
16:13 < MoonRabbit> that looks neat.
16:13 < docl> nmz787: deflecting the bolts down should shove the platforms up
16:14 < nmz787> docl: then you run out of bolts... right?
16:15 -!- strages [uid11297@gateway/web/irccloud.com/x-eyymtnnmglqtfrgd] has quit [Quit: Connection closed for inactivity]
16:16 < docl> you'd only deflect them down slightly, towards the next platform which is 'down' due to the curvature of the earth
16:16 < nmz787> but that would pass through thicker atmosphere too, then
16:17 < nmz787> guess it could work if the satellites were far enough apart
16:17 < nmz787> hard to believe/imagine successfully
16:20 -!- Jenda` [~bablbam@dekatron.hrach.eu] has quit [Ping timeout: 255 seconds]
16:27 -!- Jenda` [~bablbam@dekatron.hrach.eu] has joined ##hplusroadmap
16:40 -!- MoonRabbit [4270498f@gateway/web/freenode/ip.66.112.73.143] has left ##hplusroadmap []
16:51 < docl> the line of sight has to be far enough above the atmosphere that you can see the next platform, or you can fire the bolts slowly enough that they curve around the earth. pros and cons to either approach.
17:06 < kanzure> "Zoneout: Regularizing rnns by randomly preserving hidden activations" https://arxiv.org/pdf/1606.01305.pdf
17:07 < kanzure> "Toward storytelling from visual lifelogging: An overview" https://arxiv.org/pdf/1507.06120.pdf
17:12 < kanzure> "Generation of large-scale simulated utterances in virtual rooms to train deep-neural networks for far-field speech recognition in Google Home" https://research.google.com/pubs/pub46107.html
17:20 < kanzure> "Learning to recognize speech from chaotically synthesized data" http://web.stanford.edu/class/cs224s/reports/Faraz_Bonab.pdf   (basically: "read wikipedia and use espeak, yo")
17:36 < JayDugger>  .tw
17:36 < JayDugger> .tw https://twitter.com/kanzure
17:36 < yoleaux> Och, A cannae finde nae tweet
17:37 < JayDugger> .tw https://twitter.com/kanzure/status/894743590240387072
17:37 < yoleaux> Is there a good fMRI(or other?) measurement of "visual imagery persistence"? Might have idea for making a nootropic to improve that. (@kanzure)
17:37 < JayDugger> See 13 results at http://www.biorxiv.org/search/visual%252Bimagery%252Bpersistence
17:56 < kanzure> that's rather non-specific.
17:56 -!- mrdata_ [~mrdata@unaffiliated/mrdata] has joined ##hplusroadmap
18:00 -!- mrdata [~mrdata@unaffiliated/mrdata] has quit [Ping timeout: 276 seconds]
18:02 -!- jaboja [~jaboja@jaboja.pl] has joined ##hplusroadmap
18:02 < JayDugger> What did you want for a hundred and forty characters?
18:04 -!- danfox [~danfox@vulpinedesigns.co.uk] has quit [Ping timeout: 260 seconds]
18:05 -!- danfox [~danfox@vulpinedesigns.co.uk] has joined ##hplusroadmap
18:07 -!- preview [~quassel@2407:7000:842d:4000::3] has quit [Read error: Connection reset by peer]
18:14 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has quit [Ping timeout: 240 seconds]
18:15 -!- Qfwfq [Qfwfq@gateway/shell/suchznc/x-cqgdbkoelttxmgcy] has joined ##hplusroadmap
18:15 -!- Qfwfq [Qfwfq@gateway/shell/suchznc/x-cqgdbkoelttxmgcy] has quit [Changing host]
18:15 -!- Qfwfq [Qfwfq@unaffiliated/washirving] has joined ##hplusroadmap
18:15 -!- Qfwfq [Qfwfq@unaffiliated/washirving] has quit [Changing host]
18:15 -!- Qfwfq [Qfwfq@gateway/shell/suchznc/x-cqgdbkoelttxmgcy] has joined ##hplusroadmap
18:15 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has joined ##hplusroadmap
18:18 -!- atrus6 [~atrus6@72.241.82.247] has quit [Read error: Connection reset by peer]
18:30 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has quit [Quit: Leaving.]
18:31 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has joined ##hplusroadmap
18:34 < kanzure> "Deep compression: Compressing deep neural networks with pruning, trained quantization and huffman coding" https://arxiv.org/pdf/1510.00149.pdf (240 MB -> 6 MB models)
18:35 < kanzure> "SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5 MB model size" https://arxiv.org/pdf/1602.07360.pdf
18:41 < kanzure> adaptive apoptosis for pruning https://arxiv.org/pdf/1610.00790.pdf
18:47 < kanzure> "Rationalizing neural predictions" https://arxiv.org/pdf/1606.04155.pdf
18:48 < kanzure> "Prediction without justification has limited applicability. As a remedy, we learn to extract pieces of input text as justifications – rationales – that are tailored to be short and coherent, yet sufficient for making the same prediction. Our approach combines two modular components, generator and encoder, which are trained to operate well together. The generator specifies a distribution ...
18:48 < kanzure> ...over text fragments as candidate rationales and these are passed through the encoder for prediction. Rationales are never given during training. Instead, the model is regularized by desiderata for rationales. We evaluate the approach on multi-aspect sentiment analysis against manually annotated test cases. Our approach outperforms attention-based baseline by a significant margin. We also ...
18:49 < kanzure> ...successfully illustrate the method on the question retrieval task."
18:52 < kanzure> "Our generator specifies a distribution over possible rationales (extracted text) and the encoder maps any such text to task specific target values. They are trained jointly to minimize a cost function that favors short, concise rationales while enforcing that the rationales alone suffice for accurate prediction."
18:52 -!- streety [streety@2a01:7e00::f03c:91ff:feae:ded6] has quit [Quit: ZNC - http://znc.in]
19:13 < kanzure> recurrent network hidden state annotation, "Visual analysis of hidden state dynamics in recurrent neural networks" https://arxiv.org/pdf/1606.07461.pdf
19:32 -!- streety [streety@2a01:7e00::f03c:91ff:feae:ded6] has joined ##hplusroadmap
19:46 < kanzure> "End-to-end prediction of buffer overruns from raw source code via neural memory networks" https://arxiv.org/pdf/1703.02458.pdf
19:46 -!- yashgaroth [~yashgarot@2606:6000:cd4d:3300:f5e0:f867:a11d:8d52] has joined ##hplusroadmap
19:57 -!- jaboja [~jaboja@jaboja.pl] has quit [Remote host closed the connection]
20:23 < kanzure> "A powerful generative model using random weights for the deep image representation" http://papers.nips.cc/paper/6568-a-powerful-generative-model-using-random-weights-for-the-deep-image-representation.pdf
20:23 < kanzure> this one used random weights in their CNNs and found that the had results that were very similar to highly-trained networks. wat?
20:24 < kanzure> pasky: ^what is going on there?
20:36 < kanzure> "The convolutional neural network (CNN) will be studied in this lecture note. We introduce a rectified-correlations on a sphere (RECOS)." ok.
20:59 -!- Malvolio [~Malvolio@unaffiliated/malvolio] has quit [Quit: brb]
20:59 < kanzure> fenn: here is some anti cargo culting magic, http://diyhpl.us/~bryan/papers2/ai/machine-learning/comprehensibility/
21:08 -!- Malvolio [~Malvolio@unaffiliated/malvolio] has joined ##hplusroadmap
21:17 -!- Guest8591 [~talinck@cpe-174-97-113-184.cinci.res.rr.com] has quit [Ping timeout: 255 seconds]
21:31 < maaku> .tell aeiousomething hello received
21:31 < yoleaux> maaku: I'll pass your message to aeiousomething.
22:05 -!- justan0theruser [~justanoth@unaffiliated/justanotheruser] has joined ##hplusroadmap
22:05 -!- justan0theruser [~justanoth@unaffiliated/justanotheruser] has quit [Client Quit]
22:06 -!- justan0theruser [~justanoth@unaffiliated/justanotheruser] has joined ##hplusroadmap
22:07 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has quit [Ping timeout: 260 seconds]
22:40 -!- Malvolio [~Malvolio@unaffiliated/malvolio] has quit [Quit: Malvolio]
22:42 -!- yashgaroth [~yashgarot@2606:6000:cd4d:3300:f5e0:f867:a11d:8d52] has quit [Quit: Leaving]
22:47 -!- aeiousomething [~aeiousome@183.82.170.54] has joined ##hplusroadmap
22:56 < aeiousomething> hello
22:56 < yoleaux> 04:31Z <maaku> aeiousomething: hello received
22:56 < aeiousomething> :D
23:08 -!- CRM114 [~urchin@unaffiliated/urchin] has quit [Ping timeout: 258 seconds]
23:45 -!- augur [~augur@noisebridge130.static.monkeybrains.net] has quit [Remote host closed the connection]
--- Log closed Wed Aug 09 00:00:44 2017