|
DONT THREAD ON ME posted:there should be one where you can upload pictures of your own face and then you can jack off to porn where the male pornstars faces have bene replaced with your own if you do this you'll be a trillionaire
|
# ¿ May 29, 2019 03:29 |
|
|
# ¿ May 10, 2024 00:34 |
|
animist posted:the year is 2023. i walk into a mall kiosk labeled PORN-O-SCAN. "one scan please," i say. "i want to watch porn with my face in it." deepfaked porn where "you" are having sex with "natalie portman" is a logical step on the way to futurama-style lucy liu sex bots
|
# ¿ May 30, 2019 01:12 |
|
To me the most damning example is those papers with adversarial techniques where they apply a really subtle filter to the image and it tricks the algorithm. Like you start with an image of a turtle, and the machine recognizes it as a turtle, and then you apply a tiny convolution that affects 10% of the pixels in an almost undetectable way, and the algorithm is now certain that it's a picture of a gun. But it still looks like a turtle to you. So the takeaway is that whatever the algorithm is triggering on, it's absolutely not what humans think of as turtle-like features. It doesn't see things the way that humans do, doesn't have the same "mental" model of the world. That means you can't make any assumptions about its behavior that are based on how humans would treat a given situation. yet that's exactly what people do when they start pressing these systems into service in things like self-driving cars.
|
# ¿ May 30, 2019 06:32 |
|
DONT THREAD ON ME posted:has anyone developed an ML algorithm that can tell you if the thing you're about to post is problematic? i was thinking the other day that ML algorithms would actually be fantastic for identifying biases in a process, if anyone bothered to use them that way
|
# ¿ May 30, 2019 20:17 |
|
it would be a mitzvah if one of you would come up with an algorithm that analyses your face and tells you where to subtly apply a makeup convolution so that the system thinks your face is a turtle. tia
|
# ¿ Jun 21, 2019 18:33 |
|
Yes but curve fitting needs math while feeding data into a neural network only needs a git clone
|
# ¿ Sep 28, 2019 15:40 |
|
Captain Foo posted:is this the calculations for damage in Path of Exile no it's Loss
|
# ¿ Oct 1, 2019 23:23 |
|
Captain Foo posted:significantly slower than human reaction time so lol at the core argument for computer driving i was going to post this same thing but i looked it up and apparently the NHTSA figure for average human reaction time in an emergency situation, based on wide-ranging studies, is 1.5 to 3 seconds. three seconds. at 60mph you cover 250 feet in that time. jesus christ. start banning drivers until we get that average down under 1s please
|
# ¿ Nov 8, 2019 18:11 |
|
if elected president i will require driver re-testing every three years and drivers will be graded on a bell curve. those scoring in the lowest 25% of all drivers tested will have their license revoked.
|
# ¿ Nov 8, 2019 18:12 |
|
Schadenboner posted:See, at this point the people forced to take public transit really will be the inferiors of everyone else, though? only in driving ability, which doesn't define the person. my grandma is a wonderful person but god drat for the last few years before she gave up driving it was absolutely terrifying to be in the car with her. i know another woman who's very smart and competent in day to day life but her driving has made multiple people i know carsick to the point of vomiting. some people just aren't able to drive and that's fine. only in america is it treated as some sort of human right
|
# ¿ Nov 8, 2019 19:07 |
|
it's unambiguous in america too bc nobody in america uses permille. oh look at me using a fancy different unit that looks almost the same as one 10x larger instead of just adding a zero to the figure. this won't be confusing. like if you had millimeters and rnillimeters (100 microns), abbreviated mm and rnm 0.8% blood alcohol would kill you anyway so if that's the value you think you're talking about you know it's wrong Sagebrush fucked around with this message at 23:00 on Nov 12, 2019 |
# ¿ Nov 12, 2019 22:58 |
|
It's already Starfish (1999)
|
# ¿ Nov 17, 2019 17:29 |
|
don't mind me, just informing the director of the cornell CS department about machine learning
|
# ¿ Jan 3, 2020 03:48 |
|
yeah, uh, speaking as a college professor, that syllabus is garbage. maybe suitable for a graduate seminar where people already know what they're doing and all they need is a prompt. because that's all you're gonna get -- a prompt and "okay, uh, go make some stuff now." note that out of 9 class sessions, four are guest lecturers and three are student presentation days. perfect class for a professor who wants to get something with big name recognition on his CV and do minimal work of his own. okay class if you're already up to speed and just want a topic and a work period to fart around. poo poo class if you go into it expecting to receive some sort of useful training. e: and the two professor-led sessions are "NLP analysis" and "epidemiological modeling," both well-understood basic topics. what am quadratic curve Sagebrush fucked around with this message at 22:59 on Mar 26, 2020 |
# ¿ Mar 26, 2020 22:56 |
|
ultrafilter posted:4xx at Stanford indicates an experimental graduate level class. ah well then it's probably exactly what it says on the box. "grad students, go make a thing, if you need more information here's a list of papers to read, i'll be writing grants. if you make anything cool i'm putting my name as first author"
|
# ¿ Mar 27, 2020 00:41 |
|
DELETE CASCADE posted:what kind of hosed up PI doesn't let the main grad student on the project be first author? how else will he ever graduate
|
# ¿ Mar 27, 2020 19:04 |
|
it comes down to tenure policies and/or how machiavellian the professor in question is. some places are really lovely about granting tenure, and the only thing that matters is your authorial position on the work and the impact factor of the journal. first author on some Nature or Science papers? great, you're in. third author in some field-specific publications or, god forbid, conference presentations? hahaahaha no get the gently caress out. so in those places to save their own hides the professors will insist that they get first author credit on anything they can, which means pushing around the grad students. "you did the work in my lab so i'm automatically the first author" regardless of contribution. other places are much more realistic about how academic research works, and will weight a grad student's paper that you supervised the same as if you were the first author. this is great because you can let the student put their name first, which helps their career tremendously and is more reflective of reality anyway, knowing that your tenure committee won't just throw it away. it's much healthier and less stressful. a disadvantage of doing this is that, if you want to apply to a different university later and they subscribe to the lovely model, you might be a little screwed over unless you can convince the committee to look more deeply. so some professors still insist on being first author on everything anyway because they're relentless ladder-climbers. i was lucky enough to go to a grad school like the second example, so i had four first-authored papers when i graduated, and then got a TT position at a university that has a similar policy so i can pass it on. e: once you're a tenured professor at a cushy institution like stanford, though, there's no excuse for continuing to assign yourself first author other than ego. Sagebrush fucked around with this message at 19:45 on Mar 27, 2020 |
# ¿ Mar 27, 2020 19:41 |
|
prefabulate the amulite to reduce the chance of sinusoidal depleneration.
|
# ¿ May 10, 2020 02:40 |
|
Jabor posted:That's because you've never talked to anyone who honestly works with stats. Statistically, most statisticians are the idiots and liars
|
# ¿ Jun 17, 2020 04:50 |
|
Schadenboner posted:I think that's President Obama though? why
|
# ¿ Jun 20, 2020 20:03 |
|
ultrafilter posted:We're all in the training sets nowadays. yeah, i'd imagine if you've ever had a picture of yourself uploaded to facebook or google photos you're part of the data.
|
# ¿ Jun 21, 2020 19:04 |
|
Because furries are, in the end, always a sex thing
|
# ¿ Jun 25, 2020 19:47 |
|
hm (btw i assume everyone in yospos is worldly enough to know this, but just in case, don't put your real name in there)
|
# ¿ Jul 28, 2020 20:04 |
|
Midjack posted:big johnny cash energy https://www.youtube.com/watch?v=5cYPEkpwi-0 https://www.youtube.com/watch?v=CWtcE52LKgk
|
# ¿ Jul 28, 2020 20:22 |
|
carry on then posted:how would any of us even remember to loving breathe without the gentle guiding intellect of america's smartest boy always happy to help, carry on then
|
# ¿ Jul 28, 2020 21:03 |
|
yeah it is definitely one of the worst "AI" systems i've seen, as measured by how obvious/natural its responses are. whatever it's fitting to, it clearly isn't what humans use. john jackson: 99.9% male jill jackson: 95.8% male leslie neilsen: 70% female leslie neilson: 78% male ???
|
# ¿ Jul 29, 2020 00:10 |
|
i don't understand how they can use starcraft 2 to train an AI to drive a tank
|
# ¿ Jul 29, 2020 01:02 |
|
Shaggar posted:if you think about the modern battlefield as a collection of semi-manned drones packed full of sensors, the end result is something pretty similar to an RTS in terms of how an AI might deal with it. Midjack posted:seems like they used sc2 to teach path finding. big lols at touch screen controls in a loving tank. but i still don't understand how you "teach" from that. do you just, like, have your AI review thousands of examples of starcraft pathfinding behavior where you mark the starting point and destination and obstacles? in that case aren't you just directly ripping off blizzard's algorithm? do you use it to simulate a real-world battlefield by saying okay, a zergling is a palestinian child and a siege tank is a rocket truck, go figure it out? in that case aren't you both ripping off blizzard and using a really poor representation of your environment, full of biases? do you decompile the game and customize it to create optimized training materials for your system? in that case, why even start with starcraft? ?
|
# ¿ Jul 29, 2020 02:33 |
|
Shaggar posted:its not about actually driving the tank like first person car type stuff, its about navigation and fire control integrated into the overall battlefield. Okay but again how is StarCraft better for that than just using aerial photos of the Gaza strip?
|
# ¿ Jul 29, 2020 18:48 |
|
the IDF would totally be the guy who says 20 min nr and then does it anyway
|
# ¿ Jul 29, 2020 22:21 |
|
Pinterest Mom posted:i feel like the yom kippur war is very precisely the opposite of that how about the six-day war
|
# ¿ Jul 29, 2020 22:57 |
|
you fool! you've created a simple list of instructions for a machine to follow to slip through your net!
|
# ¿ Aug 5, 2020 23:17 |
|
the written test for a pilot's license has questions where you have to calculate weight and balance using graphs that look like this, photocopied just as poorly i.e. calculate the loading for front-seat occupants weighing 310lb, 12 gallons of fuel, and 20lb of baggage you're expected to be able to interpolate to at least half a grid square and ideally 1/10. you can't bring your own pencils in so you get a dull felt-tip pen and a plastic sheet to draw lines. the question requires multiple stages of arithmetic and the answers are multiple choice and will be like: 80.1, 80.2, 79.8. lots of people have no idea how to make a good test, apparently Sagebrush fucked around with this message at 20:30 on Sep 3, 2020 |
# ¿ Sep 3, 2020 20:28 |
|
the most horrifying part of the minority report universe is the nightstick that makes you barf imagine the damage a 13 year old could do with one of those
|
# ¿ Sep 5, 2020 03:31 |
|
is the joke here that Big Data is the discipline formerly known as statistics
|
# ¿ Sep 23, 2020 20:07 |
|
im in awe at the stupidity of this model https://twitter.com/baumard_nicolas/status/1308715661154308103 happy smiling man determined by (points at seagull) to be more trustworthy than dour looking angry man! what happens if they run it with trump doing that smarmy grin he does and joe looking scared and confused
|
# ¿ Sep 25, 2020 20:46 |
|
jehanne butler was right
|
# ¿ Sep 26, 2020 20:24 |
|
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 |
|
we can optimize the solution by removing all these extraneous handlers. bonus time
|
# ¿ Jan 3, 2021 05:57 |
|
|
# ¿ May 10, 2024 00:34 |
|
this is the current state of the art in image recognition
|
# ¿ Mar 31, 2021 06:12 |