--- Log opened Thu Aug 10 00:00:45 2017 00:06 -!- esmerelda [~mabel@unaffiliated/jacco] has quit [Quit: WeeChat 1.8] 00:06 -!- esmerelda [~mabel@174-24-228-29.tukw.qwest.net] has joined ##hplusroadmap 00:06 -!- esmerelda [~mabel@174-24-228-29.tukw.qwest.net] has quit [Changing host] 00:06 -!- esmerelda [~mabel@unaffiliated/jacco] has joined ##hplusroadmap 00:15 -!- helleshin [~talinck@cpe-174-97-113-184.cinci.res.rr.com] has joined ##hplusroadmap 00:39 -!- aeiousomething [~aeiousome@183.82.170.54] has joined ##hplusroadmap 02:13 -!- augur [~augur@162.245.20.162] has joined ##hplusroadmap 02:14 -!- CRM114 [~urchin@unaffiliated/urchin] has joined ##hplusroadmap 02:17 -!- augur [~augur@162.245.20.162] has quit [Ping timeout: 248 seconds] 02:41 -!- CheckDavid [uid14990@gateway/web/irccloud.com/x-mxoaldetjklhyjan] has joined ##hplusroadmap 03:40 -!- bluebear_ [~dluhos@80.95.97.194] has joined ##hplusroadmap 04:14 -!- jaboja [~jaboja@jaboja.pl] has joined ##hplusroadmap 04:33 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has joined ##hplusroadmap 04:50 -!- CheckDavid [uid14990@gateway/web/irccloud.com/x-mxoaldetjklhyjan] has quit [Quit: Connection closed for inactivity] 05:11 -!- CRM114 [~urchin@unaffiliated/urchin] has quit [Remote host closed the connection] 05:36 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has quit [Ping timeout: 255 seconds] 05:48 < kanzure> "Node-by-node greedy deep learning for interpretable features" http://diyhpl.us/~bryan/papers2/ai/machine-learning/comprehensibility/Node-by-node%20greedy%20deep%20learning%20for%20interpretable%20features%20-%202016.pdf (2016) 05:52 -!- CRM114 [~urchin@unaffiliated/urchin] has joined ##hplusroadmap 06:20 -!- jaboja [~jaboja@jaboja.pl] has quit [Ping timeout: 240 seconds] 06:26 < kanzure> .title http://www.nss.org/settlement/nasa/Nowicki/SPBI110.HTM 06:26 < yoleaux> Ice Gun 06:27 < kanzure> "The ice gun is a barbell-shaped ramjet flying in a tunnel cut through antarctic ice sheet and filled with hydrogen at about twice the atmospheric pressure. Forward half of the ramjet is the umbrella projectile carrying cargo, while the aft half carries liquid oxygen. Upon leaving the gun, the aft half separates and disintegrates in the atmosphere. To reduce the stress of acceleration force ... 06:28 < kanzure> ...on the cargo, the tunnel is about 1000 km long and the projectile employs hydrogen injection. The ice gun belongs to the transverse gas gun family because the hydrogen flow is transverse to the projectile direction. The maximum velocity is about 8 km/s. There is no bibliography. The minimum mass is only 100 tons not counting the ice. The ice gun stands out due to its low cost (less than 1% of ... 06:28 < kanzure> ...the cost of chemical rocket launchers) and the ability to carry fragile cargo and people. Its only shortcoming is the reliance on the ice sheets which are melting due to global warming. A short version of the gun also deserves attention. For example, a 100 km long gun accelerating projectiles to 4.4 km/s would expose the cargo to the acceleration of 10 G, and would be one order of magnitude ... 06:28 < kanzure> ...more economical than the chemical rocket launchers." 06:47 -!- jaboja [~jaboja@jaboja.pl] has joined ##hplusroadmap 06:49 < kanzure> .tw https://twitter.com/jjvincent/status/895554642133659648 06:49 < yoleaux> the most cyberpunk news you will read today https://www.wired.com/story/malware-dna-hack/ https://pbs.twimg.com/media/DG2mIOOW0AAoe7P.jpg (@jjvincent) 06:49 < kanzure> .tw https://twitter.com/kanzure/status/895642980085661697 06:49 < yoleaux> @jjvincent these guys are misleading if they are saying this: "Needless to say, any possible DNA-based hacking is years away" (@kanzure, in reply to tw:895554642133659648) 06:53 -!- Storyteller [~Storytell@unaffiliated/storyteller] has joined ##hplusroadmap 07:05 -!- justan0theruser [~justanoth@unaffiliated/justanotheruser] has quit [Ping timeout: 260 seconds] 07:20 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has joined ##hplusroadmap 07:37 -!- jaboja [~jaboja@jaboja.pl] has quit [Ping timeout: 276 seconds] 07:43 < fenn> years away, as in the conference after next 07:48 < kanzure> right, have to adjust for the internet time dilation factor 07:49 < kanzure> .tw https://twitter.com/bobcrunkhouse/status/895583026461028352 07:49 < yoleaux> @jjvincent oh god it's literally the terrible scene in bones where a computer explodes cos a corpse uploaded "malware fractals" carved into the bones. (@bobcrunkhouse, in reply to tw:895554642133659648) 07:49 < kanzure> .tw https://twitter.com/DarrenCHoffman/status/895642634768728064 07:49 < yoleaux> @jjvincent @ebassi yeah, that's cute, but the DNA is just an input string at that point, no different than any other buffer overflow or similar type of exploit (@DarrenCHoffman, in reply to tw:895554642133659648) 07:58 -!- jaboja [~jaboja@jaboja.pl] has joined ##hplusroadmap 08:06 -!- jaboja [~jaboja@jaboja.pl] has quit [Ping timeout: 240 seconds] 08:06 -!- jaboja [~jaboja@jaboja.pl] has joined ##hplusroadmap 08:19 < kanzure> "Compromising computers with synthesized DNA" http://dnasec.cs.washington.edu/dnasec.pdf https://news.ycombinator.com/item?id=14979120 08:22 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has quit [Ping timeout: 240 seconds] 08:22 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has joined ##hplusroadmap 08:26 -!- jaboja [~jaboja@jaboja.pl] has quit [Ping timeout: 248 seconds] 08:28 < kanzure> .wik KL-VS 08:28 < yoleaux> "KLVS (107.3 FM) is a radio station broadcasting a Contemporary Christian music format from K-LOVE, licensed to Livermore, California, United States. The station is owned by San Joaquin Broadcasting Company." — https://en.wikipedia.org/wiki/KLVS 08:28 < kanzure> hmm... not quite what i wanted. 08:28 < kanzure> .g Klotho KL-VS polymorphism 08:28 < yoleaux> https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5004824/ 08:28 < kanzure> .title 08:28 < yoleaux> Lack of functional KL-VS polymorphism of the KLOTHO gene in the Korean population 08:29 < kanzure> "For her part, Dubal is interested in partnering with biotech and pharma companies with a view to moving toward human trials of klotho itself" 08:31 < kanzure> "Frameworks for proof-of-concept clinical trials of interventions that target fundamental aging processes" https://www.academic.oup.com/biomedgerontology/article-lookup/doi/10.1093/gerona/glw126 08:36 -!- augur [~augur@162.245.20.162] has joined ##hplusroadmap 09:09 -!- Storyteller [~Storytell@unaffiliated/storyteller] has quit [Ping timeout: 240 seconds] 09:24 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has quit [Ping timeout: 240 seconds] 09:25 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has joined ##hplusroadmap 09:26 -!- sachy [~sachy@nat.brmlab.cz] has joined ##hplusroadmap 09:33 -!- jaboja [~jaboja@jaboja.pl] has joined ##hplusroadmap 10:06 -!- augur [~augur@162.245.20.162] has quit [Remote host closed the connection] 10:27 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has quit [Ping timeout: 255 seconds] 10:28 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has joined ##hplusroadmap 10:28 -!- augur [~augur@162.245.20.162] has joined ##hplusroadmap 10:33 -!- augur [~augur@162.245.20.162] has quit [Ping timeout: 240 seconds] 10:53 -!- c0rw1n [~c0rw1n@91.177.138.19] has quit [Ping timeout: 248 seconds] 10:54 -!- juri_ [~juri@205.166.94.162] has quit [Ping timeout: 240 seconds] 10:57 -!- Solgriffin [~Sol@c-69-141-24-242.hsd1.nj.comcast.net] has joined ##hplusroadmap 11:02 -!- juri_ [~juri@205.166.94.162] has joined ##hplusroadmap 11:21 < Jenda`> kanzure: some more info about "non-linear random projections" - https://arxiv.org/abs/1412.8307 http://www.sciencedirect.com/science/article/pii/S0925231206000385 https://www.reddit.com/r/MachineLearning/comments/34u0go/yann_lecun_whats_so_great_about_extreme_learning/ http://www.ntu.edu.sg/home/egbhuang/ 11:23 < kanzure> .title 11:23 < yoleaux> Extreme Learning Machines: Random Neurons, Random Features, Kernels 11:24 < Jenda`> tl;dr you generate ~1000 random projections with ReLU and then least-square fit it with your training data 11:25 < Jenda`> (I think that sciencedirect paper is the best introduction; it's ons sci-hub) 11:30 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has quit [Ping timeout: 255 seconds] 11:36 -!- atrus6 [~atrus6@72.241.82.247] has quit [Quit: Leaving] 11:42 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has joined ##hplusroadmap 11:57 < fltrz> "The main innovation is to ensure each hidden-unit operates only on a randomly sized and positioned patch of each image." 11:57 < kanzure> and what do other hidden units in other designs do? 11:58 < fltrz> I believe "features" are in fact kinds of symmetries 11:58 < fltrz> we think we are training dog features, while in fact we train symmetries, like a round eye has rotational symmetry 11:59 < kanzure> you don't think that's a dog feature? 11:59 < fltrz> animals can direct their eyes at different parts of the view, effictively shifting -> translation symmetry 12:00 < fltrz> coming closer or farther is scaling symmetry 12:02 < fltrz> we think we train on 100 dog pictures, but they are from different distances, so in fact we expose symmetries (and their corresponding invariants) which would just as well show up with random images 12:02 < fltrz> training on 100 car pictures will also reveal these symmetries in natural images 12:03 < kanzure> it's true that two round circles should be marked as possibly a dog face eyeballs, but there are also many other things it should be marked as possibly being, long before other information is interpreted 12:03 < fltrz> sure there will be many symmetries 12:04 < fltrz> the local hair direction, has local translation symmetry 12:05 < fltrz> without using convolution operation, we could have a identically weighted neurons at each location in an image, and it makes the network large and redundant (the way it is in our brain because the information has to be processed in parallel) 12:06 < fltrz> but the convolutional neural networks store the set of filters once, and we use a convolution 12:06 < fltrz> which drastically reduces the number of neurons for identical mathematical result 12:06 < fltrz> thats because we recognize the translation symmetry and understand fourier transforms 12:08 < fltrz> for scaling symmetry, consider the fourier transform in music, the same waveform played at different frequencies show up as scaled patterns in frequency domain 12:08 < fltrz> but with a constant q transform, no matter at what frequency a waveform is played at, it will have the same intensity profile, only shifted across the spectrum (NOT scaled) 12:09 < fltrz> so the constant q transform is much more usefull from a signal processing standpoint 12:10 < fltrz> (i.e. filter/correlator that detects say a square wave in a constant q transform spectrum is independent of frequency, this is not the case for fourier transform 12:12 < fltrz> for images a 2D constant q transform (which I find no literature on, no-one seems to define a 2D constant q transform nor its application in images) 12:12 < fltrz> then the same filter could recognize say a specific face independent of scaling (camera closer or further from face) 12:15 < fltrz> (instead of needing a set of filters for each scale) 12:16 -!- c0rw1n [~c0rw1n@tmo-110-77.customers.d1-online.com] has joined ##hplusroadmap 12:16 < fltrz> a bit old but http://academics.wellesley.edu/Physics/brown/pubs/cq1stPaper.pdf 12:17 < fltrz> so I propose to train NN's on 2d constant q transformed images 12:18 < kanzure> have i mentioned my deepspeech implementation 12:19 < fltrz> no, not in my presence 12:19 -!- hehelleshin [~talinck@cpe-174-97-113-184.cinci.res.rr.com] has joined ##hplusroadmap 12:19 < kanzure> i converted baidu's version to tensorflow/keras 12:21 -!- c0rw1n [~c0rw1n@tmo-110-77.customers.d1-online.com] has quit [Ping timeout: 255 seconds] 12:23 -!- helleshin [~talinck@cpe-174-97-113-184.cinci.res.rr.com] has quit [Ping timeout: 240 seconds] 12:31 -!- c0rw1n [~c0rw1n@ptr-178-50-68-243.dyn.mobistar.be] has joined ##hplusroadmap 12:36 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has quit [Ping timeout: 248 seconds] 12:50 -!- Cory [~Cory@unaffiliated/cory] has quit [Ping timeout: 255 seconds] 12:58 -!- Pasha [~Cory@unaffiliated/cory] has joined ##hplusroadmap 12:58 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has joined ##hplusroadmap 13:00 -!- Pasha is now known as Cory 13:01 -!- poppingtonic1 [~brian@51.15.130.225] has joined ##hplusroadmap 13:02 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has quit [Remote host closed the connection] 13:03 -!- atrus6 [~atrus6@72.241.82.247] has joined ##hplusroadmap 13:18 -!- c0rw1n [~c0rw1n@ptr-178-50-68-243.dyn.mobistar.be] has quit [Ping timeout: 260 seconds] 13:35 -!- c0rw1n [~c0rw1n@ptr-178-50-68-243.dyn.mobistar.be] has joined ##hplusroadmap 13:40 -!- bluebear_ [~dluhos@80.95.97.194] has quit [Quit: Leaving.] 13:46 -!- Storyteller [~Storytell@unaffiliated/storyteller] has joined ##hplusroadmap 14:17 -!- aeiousomething [~aeiousome@183.82.170.54] has left ##hplusroadmap ["baaai"] 15:17 -!- Storyteller [~Storytell@unaffiliated/storyteller] has quit [Read error: Connection timed out] 15:17 -!- Storyteller [~Storytell@unaffiliated/storyteller] has joined ##hplusroadmap 15:33 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has joined ##hplusroadmap 15:48 -!- c0rw1n [~c0rw1n@ptr-178-50-68-243.dyn.mobistar.be] has quit [Ping timeout: 260 seconds] 15:49 -!- augur [~augur@noisebridge130.static.monkeybrains.net] has joined ##hplusroadmap 15:50 -!- c0rw1n [~c0rw1n@ptr-178-50-84-163.dyn.mobistar.be] has joined ##hplusroadmap 16:00 -!- c0rw1n [~c0rw1n@ptr-178-50-84-163.dyn.mobistar.be] has quit [Read error: error:1408F10B:SSL routines:SSL3_GET_RECORD:wrong version number] 16:01 -!- c0rw1n [~c0rw1n@ptr-178-50-84-163.dyn.mobistar.be] has joined ##hplusroadmap 16:04 -!- poppingtonic1 [~brian@51.15.130.225] has quit [Ping timeout: 260 seconds] 16:14 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has joined ##hplusroadmap 16:17 -!- atrus6 [~atrus6@72.241.82.247] has quit [Read error: Connection reset by peer] 16:20 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has quit [Remote host closed the connection] 16:20 -!- c0rw1n [~c0rw1n@ptr-178-50-84-163.dyn.mobistar.be] has quit [Ping timeout: 255 seconds] 16:29 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has quit [Ping timeout: 260 seconds] 16:29 -!- c0rw1n [~c0rw1n@ptr-178-50-84-163.dyn.mobistar.be] has joined ##hplusroadmap 16:35 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has joined ##hplusroadmap 16:41 < kanzure> the future https://i.imgur.com/eROC0r9_d.jpg 16:47 < nmz787> hehe, if my wife would just get her driving license... 16:52 -!- Gurkenglas [~Gurkengla@dslb-178-000-090-041.178.000.pools.vodafone-ip.de] has joined ##hplusroadmap 16:55 < kanzure> hmm bad image 16:56 < kanzure> the future http://i.imgur.com/eROC0r9.jpg 16:56 < kanzure> there we go. 17:02 < nmz787> I totally thought he was cooking 17:03 < nmz787> (food) 17:07 < kanzure> is he not? 17:08 < kanzure> or is he instead operating a blade saw on the highway? 17:26 -!- c0rw1n [~c0rw1n@ptr-178-50-84-163.dyn.mobistar.be] has quit [Ping timeout: 276 seconds] 17:48 -!- jaboja [~jaboja@jaboja.pl] has quit [Remote host closed the connection] 18:03 -!- danfox [~danfox@vulpinedesigns.co.uk] has quit [Ping timeout: 260 seconds] 18:06 -!- danfox [~danfox@vulpinedesigns.co.uk] has joined ##hplusroadmap 18:07 -!- Storyteller [~Storytell@unaffiliated/storyteller] has quit [Ping timeout: 276 seconds] 18:17 -!- Storyteller [~Storytell@unaffiliated/storyteller] has joined ##hplusroadmap 18:29 -!- mrdata [~mrdata@unaffiliated/mrdata] has joined ##hplusroadmap 18:30 -!- c0rw1n [~c0rw1n@ptr-178-50-84-163.dyn.mobistar.be] has joined ##hplusroadmap 18:42 < JayDugger> Want to predict when someone dies using power tools in a self-driving car? 18:43 -!- JayDugger [~jwdugger@47.185.237.246] has quit [Quit: Leaving.] 18:47 -!- c0rw1n [~c0rw1n@ptr-178-50-84-163.dyn.mobistar.be] has quit [Ping timeout: 255 seconds] 18:47 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has quit [Read error: Connection reset by peer] 18:48 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has joined ##hplusroadmap 18:51 -!- Storyteller [~Storytell@unaffiliated/storyteller] has quit [Quit: Leaving] 18:58 -!- mrdata- [~mrdata@unaffiliated/mrdata] has joined ##hplusroadmap 19:01 -!- c0rw1n [~c0rw1n@ptr-178-50-69-17.dyn.mobistar.be] has joined ##hplusroadmap 19:01 -!- mrdata [~mrdata@unaffiliated/mrdata] has quit [Ping timeout: 240 seconds] 19:01 -!- c0rw1n [~c0rw1n@ptr-178-50-69-17.dyn.mobistar.be] has quit [Remote host closed the connection] 19:02 -!- c0rw1n [~c0rw1n@ptr-178-50-69-17.dyn.mobistar.be] has joined ##hplusroadmap 19:03 -!- preview [~quassel@2407:7000:842d:4000::3] has quit [Read error: Connection reset by peer] 19:05 -!- preview [~quassel@2407:7000:842d:4000::3] has joined ##hplusroadmap 19:14 -!- c0rw1n [~c0rw1n@ptr-178-50-69-17.dyn.mobistar.be] has quit [Ping timeout: 255 seconds] 19:17 -!- Sol [~Sol@c-69-141-24-242.hsd1.nj.comcast.net] has joined ##hplusroadmap 19:18 -!- sivoais [~zaki@unaffiliated/sivoais] has quit [Ping timeout: 260 seconds] 19:19 -!- sivoais [~zaki@unaffiliated/sivoais] has joined ##hplusroadmap 19:22 -!- yashgaroth [~yashgarot@2606:6000:cd4d:3300:f5e0:f867:a11d:8d52] has joined ##hplusroadmap 19:28 -!- Solgriffin [~Sol@c-69-141-24-242.hsd1.nj.comcast.net] has quit [Ping timeout: 260 seconds] 19:30 -!- balrog [~balrog@unaffiliated/balrog] has quit [Ping timeout: 240 seconds] 19:32 -!- maaku [~mark@173.234.25.100] has quit [Quit: Lost terminal] 19:33 -!- balrog [~balrog@unaffiliated/balrog] has joined ##hplusroadmap 19:35 -!- maaku [~mark@173.234.25.100] has joined ##hplusroadmap 19:39 -!- c0rw1n [~c0rw1n@ptr-178-50-78-113.dyn.mobistar.be] has joined ##hplusroadmap 19:59 < kanzure> iej01hadfaljafa[p0ehqut823q8dhfsalfdas 20:06 -!- Gurkenglas [~Gurkengla@dslb-178-000-090-041.178.000.pools.vodafone-ip.de] has quit [Ping timeout: 240 seconds] 20:06 < mrdata-> hello kitty 20:07 < ybit> kanzure, check your last message 20:07 < ybit> missed this one: https://www.forbes.com/sites/jenniferhicks/2016/03/22/fda-approved-3d-printed-drug-available-in-the-us/#6dcebe87666b 20:07 < kanzure> my last who? 20:08 < ybit> 19:59 < kanzure> iej01hadfaljafa[p0ehqut823q8dhfsalfdas 20:11 -!- Gurkenglas [~Gurkengla@dslb-178-000-090-041.178.000.pools.vodafone-ip.de] has joined ##hplusroadmap 20:12 < kanzure> yeah it's a good one 20:12 -!- Sol is now known as Solgriffin 20:13 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has quit [Ping timeout: 240 seconds] 20:15 -!- c0rw1n [~c0rw1n@ptr-178-50-78-113.dyn.mobistar.be] has quit [Ping timeout: 240 seconds] 20:18 -!- c0rw1n [~c0rw1n@ptr-178-50-78-145.dyn.mobistar.be] has joined ##hplusroadmap 20:38 -!- mrdata- [~mrdata@unaffiliated/mrdata] has quit [Read error: Connection reset by peer] 20:42 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has joined ##hplusroadmap 20:51 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has quit [Remote host closed the connection] 21:00 -!- c0rw1n [~c0rw1n@ptr-178-50-78-145.dyn.mobistar.be] has quit [Ping timeout: 260 seconds] 21:06 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has joined ##hplusroadmap 21:13 < docl> kanzure: what did you think of the ice gun? it reminded me a bit of a conversation we were having a while back about using icebergs for something 21:24 < docl> hmm. seawater does freeze if you cool it enough. so you could in principle have a device that manufactures a tube shaped ice structure as it moves along. with fiber supports, maybe it could get long enough to be useful without tending to snap in the ocean currents. this would be useful for artificial roadways and such as well. 21:27 -!- Gurkenglas [~Gurkengla@dslb-178-000-090-041.178.000.pools.vodafone-ip.de] has quit [Ping timeout: 276 seconds] 21:41 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has joined ##hplusroadmap 22:47 -!- yashgaroth [~yashgarot@2606:6000:cd4d:3300:f5e0:f867:a11d:8d52] has quit [Quit: Leaving] 23:04 -!- poppingtonic [~brian@unaffiliated/poppingtonic] has quit [Ping timeout: 240 seconds] --- Log closed Fri Aug 11 00:00:46 2017