|
echinopsis posted:hmm thanks yeah the style transfer stuff is really loving hard to get good results i tried to make some 2020 food lines photos look like norman rockwell and these are probably my best results after generating 100+ (and learning about how to work around interactivity requirements on colab) you might try very simplistic images that mostly have only a texture or only a pattern or just one style so the ml can transfer just one thing, won't produce as "realstic" results but would probably look cool
|
# ? Oct 29, 2020 19:13 |
|
|
# ? Apr 26, 2024 21:01 |
|
hmm yeah: they all just look like style transfer. tempted to blend back with the original so effect is more subtle but yeah I was thinking i could maybe introduce a style somehow so my renders would look less rendered and more drawn. but coming to conclusion this isn’t the way
|
# ? Oct 30, 2020 00:49 |
|
op have you tried the prisma app
|
# ? Oct 30, 2020 01:42 |
|
is that the one that does more pixel based effects rather than anything “neural”? kinda like loving around with pixels but not sure the best way
|
# ? Oct 30, 2020 02:00 |
|
https://twitter.com/seagull81/status/1320132156774023168
|
# ? Oct 30, 2020 04:03 |
|
so like you know in star trek all the doors on the enterprise slide open, but it was just some person pulling on wires or something at the right time in the script, not a real automatic door, so it appears to work not on motion, but like a person's intent because sometimes there'd be a whole conversation right next to a door and it'll magically open when riker is done chatting up troi or whatever ugh. could you feed in all the ridiculous footage into all this machine learning poo poo and end up with software that'd actually open a door based on intent just like how it appears on the show I don't know
|
# ? Oct 30, 2020 16:55 |
|
fack you posted:so like you know in star trek all the doors on the enterprise slide open, but it was just some person pulling on wires or something at the right time in the script, not a real automatic door, so it appears to work not on motion, but like a person's intent because sometimes there'd be a whole conversation right next to a door and it'll magically open when riker is done chatting up troi or whatever ugh. could you feed in all the ridiculous footage into all this machine learning poo poo and end up with software that'd actually open a door based on intent just like how it appears on the show I don't know Sure, as long as you define ML and AI as "paying 3rd world people pennies to watch video and push button when ready".
|
# ? Oct 30, 2020 16:57 |
|
fack you posted:so like you know in star trek all the doors on the enterprise slide open, but it was just some person pulling on wires or something at the right time in the script, not a real automatic door, so it appears to work not on motion, but like a person's intent because sometimes there'd be a whole conversation right next to a door and it'll magically open when riker is done chatting up troi or whatever ugh. could you feed in all the ridiculous footage into all this machine learning poo poo and end up with software that'd actually open a door based on intent just like how it appears on the show I don't know i think this is one of those cases which some ml model would do pretty well yeah. not like fully star trek magical, but differentiating between someone going in, someone standing still, and someone walking past should be doable in a fairly accurate way. and is actually not a bad idea, the amount of energy lost to automatic doors opening unnecessarily in winters and such probably adds up considering the relative simplicity of implementation.
|
# ? Oct 30, 2020 17:31 |
|
all the people on star trek have universal translator brain implants so maybe they actually do read door-opening intent and broadcast it to the door. they also never pause in front of the door to let it open, just stride straight through and expect it to be ready as they get there, so i wonder how often people in star trek show up with a broken nose because the doors are malfunctioning
|
# ? Oct 30, 2020 17:46 |
|
lmao there's no way that would work irl without false negatives/positives you could try converting all doors simultaneously while educating everyone on the new door paradigm, but still there'd be people with loopy strides or someone forgetting about the new doors & hesitating or whatever even humans employed full time as doormen cant tell correctly 100%
|
# ? Oct 30, 2020 18:46 |
|
its in their comm badge with a connection back to the ship for the heavy lifting. they use a lost connection to the ship as a mcguffin lots of times i imagine it as alexaprise is peeping in on all the conversations and collecting motion data from the badge since it does idk some health stuff probably, then figures out youre done getting burned by data as you leave engineering and opens the door in other words bezos will be selling intent-based doors for home deliveries in 2023. also intent-based restocking as you reach for the second last cola in 2022
|
# ? Oct 30, 2020 18:52 |
|
it would be like one of those comedy bits where two people keep starting a sentence at the same time and say no sorry what were you saying
|
# ? Oct 30, 2020 18:53 |
|
somebody post that gif of the seagull tricking the convenience store door sensor and running back out with a bag of chips or maybe it was a duck I forget
|
# ? Nov 7, 2020 14:16 |
|
Feisty-Cadaver posted:somebody post that gif of the seagull tricking the convenience store door sensor and running back out with a bag of chips https://www.youtube.com/watch?v=Kqy9hxhUxK0
|
# ? Nov 8, 2020 08:03 |
|
Source
|
# ? Nov 13, 2020 05:24 |
|
https://www.wsj.com/articles/when-the-machines-learn-to-price-gouge-11601281879?redirect=amp#click=https://t.co/BDhFcxs8fsquote:In such markets where both stations appeared to adopt algorithmic software, as estimated by sudden changes in the size and rapidity of price changes, margins increased by an average of almost 30%. Without pricing software at both stations, margins were unchanged. Notably this would be illegal of people were doing it - the computers presumably manage some sort of weird pseudo-communication through their price signals and end up forming a cartel!
|
# ? Nov 16, 2020 11:36 |
|
BYOB has a thread where people visit threads from other forums and write a trip report. Someone got assigned this thread, and here's what they wrote:beer pal posted:u know how when you have a car you're all like yeah i get it... the gas blows up and that sends the pistons moving and then the wheels turn or whatever.. and then you take it to the mechanic and they say your screwjank is all rusted and your double boiler has got an oil leak int he drivecrank.. well thats what i got from the op of this thread. everybody knows what machine learning is its when the machines learn. even with the probably laymans explanation and sardonic tone all the computer words got me messed up. i read one computer word and its lights off up there. but anyway after that its people making small joke posts much like a yobbo would do on the subject of machine learning. This person gets it.
|
# ? Nov 16, 2020 16:44 |
|
how much is jeffrey paying them for that dreary review work?
|
# ? Nov 16, 2020 17:01 |
|
Cybernetic Vermin posted:how much is jeffrey paying them for that dreary review work? well it is BYOB, so existence in itself is the reward.
|
# ? Nov 16, 2020 17:04 |
|
beer pal seems to have a solid understanding of machine learning, missing only this key image that i guess i probably got from this thread but i'm too lazy to find the post to quote
|
# ? Nov 16, 2020 17:09 |
|
big scary monsters posted:beer pal seems to have a solid understanding of machine learning, missing only this key image that i guess i probably got from this thread but i'm too lazy to find the post to quote it's in the op, op
|
# ? Nov 16, 2020 17:31 |
|
i ran it throguh a nn and now it's bigger and more true
|
# ? Nov 16, 2020 17:39 |
|
Trabisnikof posted:Nature Communications charges a $5,700 publishing fee too lol Nature Comms is basically a scam. It's PLoS ONE with better brand recognition and faux selectivity for papers that aren't actually of general interest (Nature), which is apparently worth a 4x premium over just publishing in PLoS ONE or knockoff PLoS ONE (Scientific Reports). I guess if your tenure committee is full of idiots who only care about whether your publication record has lots of high impact factor numbers and contains the word "nature" it's worth it.
|
# ? Nov 16, 2020 22:30 |
|
maybe i should edit the OP to have less words. maybe just like "algorithms are out to get you." and that picture
|
# ? Nov 18, 2020 07:49 |
|
oh also, this short story got posted in the c-spam doomsday econ thread (i think??) and its really good. pretty solid examination of "but what if the machines turn on us???": https://rifters.com/real/shorts/PeterWatts_Malak.pdf
|
# ? Nov 18, 2020 08:01 |
|
a surprisingly optimistic ending for watts
|
# ? Nov 18, 2020 11:38 |
|
why? do some of the humans survive?
|
# ? Nov 18, 2020 12:21 |
|
fart simpson posted:why? do some of the humans survive? no
|
# ? Nov 18, 2020 12:27 |
|
animist posted:oh also, this short story got posted in the c-spam doomsday econ thread (i think??) and its really good. pretty solid examination of "but what if the machines turn on us???": https://rifters.com/real/shorts/PeterWatts_Malak.pdf Yeah this is cool because it sort of gets to the center of "how innocuous rules can cause unintended behaviour" which, at least for the near future is a much bigger concern than the old sci-fi AI plot of a machine attaining full sentience and deciding that people cause bad stuff like ecological collapse and deciding to purge us. I always like the "paperclip maximiser" thing for that too.
|
# ? Nov 18, 2020 12:46 |
|
https://twitter.com/xkcd/status/1330309742535827467
|
# ? Nov 22, 2020 04:20 |
|
Imagine having quoted a Randalltweet.
|
# ? Nov 22, 2020 15:55 |
|
https://twitter.com/albrgr/status/1333838686992044032
|
# ? Dec 1, 2020 21:25 |
|
artificial stupidity
|
# ? Dec 1, 2020 21:38 |
|
you'd think that 20 years and counting of seo would have taught people the value of manually gaming a "smart" system
|
# ? Dec 1, 2020 21:49 |
|
https://aws.amazon.com/blogs/aws/amazon-devops-guru-machine-learning-powered-service-identifies-application-errors-and-fixes/ ah yes this will go well
|
# ? Dec 1, 2020 21:51 |
|
source, via calling bullshit
|
# ? Dec 6, 2020 07:39 |
|
https://twitter.com/rajiinio/status/1343278077229588481
|
# ? Dec 28, 2020 01:27 |
|
https://twitter.com/DeepCapybara/status/1344840846177427457
|
# ? Jan 3, 2021 05:33 |
|
yeah. it actually works out for all the things he puts in the square hole during training, so the problem is solved for all inputs in perpetuity
|
# ? Jan 3, 2021 05:48 |
|
|
# ? Apr 26, 2024 21:01 |
|
we can optimize the solution by removing all these extraneous handlers. bonus time
|
# ? Jan 3, 2021 05:57 |