Slashdot Mirror
Researchers Are Training Image-Generating AI With Fewer Labels (venturebeat.com)
An anonymous reader shares a report: Generative AI models have a propensity for learning complex data distributions, which is why they're great at producing human-like speech and convincing images of burgers and faces. But training these models requires lots of labeled data, and depending on the task at hand, the necessary corpora are sometimes in short supply.
The solution might lie in an approach proposed by researchers at Google and ETH Zurich. In a paper [PDF] published on the preprint server Arxiv.org ("High-Fidelity Image Generation With Fewer Labels"), they describe a "semantic extractor" that can pull out features from training data, along with methods of inferring labels for an entire training set from a small subset of labeled images. These self- and semi-supervised techniques together, they say, can outperform state-of-the-art methods on popular benchmarks like ImageNet.
"In a nutshell, instead of providing hand-annotated ground truth labels for real images to the discriminator, we ... provide inferred ones," the paper's authors explained. In one of several unsupervised methods the researchers posit, they first extract a feature representation -- a set of techniques for automatically discovering the representations needed for raw data classification -- on a target training dataset using the aforementioned feature extractor.
The solution might lie in an approach proposed by researchers at Google and ETH Zurich. In a paper [PDF] published on the preprint server Arxiv.org ("High-Fidelity Image Generation With Fewer Labels"), they describe a "semantic extractor" that can pull out features from training data, along with methods of inferring labels for an entire training set from a small subset of labeled images. These self- and semi-supervised techniques together, they say, can outperform state-of-the-art methods on popular benchmarks like ImageNet.
"In a nutshell, instead of providing hand-annotated ground truth labels for real images to the discriminator, we ... provide inferred ones," the paper's authors explained. In one of several unsupervised methods the researchers posit, they first extract a feature representation -- a set of techniques for automatically discovering the representations needed for raw data classification -- on a target training dataset using the aforementioned feature extractor.
So basically n dimensional (or really just n instances of) Newtons method? Sounds like a good place to get sound bites for a trump rally.
Oh FFS! Give it a rest- stop spamming the forum with your personal vendetta against her.
They are bouncing to the left and to the right. Its my belief that my DAMN balls should be held every night
When her propaganda stops, the anti-propaganda stops. #Rope is coming.
Researchers Are Training Image-Generating AI...
Story:
The solution might lie in an approach proposed by researchers at Google and ETH Zurich...
The researchers aren't training anything. They just hypothesized that it might be possible to use AI to train AI. Then their heads exploded.
Nothing to it
When the whole fiasco of questionable suggestions cropped up with Youtube recently, I thought that this was probably exactly what was already being done: get a bunch of videos with labels, train machine learning that they're alike, track people who like them, show other videos by the same people, if the same people like them as well and don't seem to like videos with other labels then add them to your labelled group as the same thing. Obviously, that's a recipe for disaster.
Who would have thought. Oh, right, I learned that about 30 years ago at university in my CS studies.
Most ACs are not even worth the keystrokes to insult them. Be generically insulted by this and ignored otherwise.
So we're training machine learning algorithms with data that was generated by machine learning algorithms?
And we're using those algorithms in situations where we didn't have much data, which may often mean they are complex situations?
This sounds like a bias-factory, breaking some kind of law of entropy.
My comprehension regressed upon encountering this sentence.
That is one computer guessing what something is so another computer can learn to identify it?!?!?
I think someone skipped their logic class one too many times :O
Those artificially generated fire trucks sure are funky looking. They immediately stand out as fire-trucks, but as you look more closely, they have weird details in weird spots, and duplicate things that shouldn't be duplicated in practice. iLSD or a transporter accident.
Table-ized A.I.
Is "AI" more than just glorified approximation or curve fitting?
If not its all very ridiculous - even the name itself is marketing bullshit. At least they call it "machine learning" in the paper unlike stupid slashot.
This type of "AI" is really not more. Non-statistical approaches are different, but about as "intelligent".
Most ACs are not even worth the keystrokes to insult them. Be generically insulted by this and ignored otherwise.