Exactly. But not just about you, about anything. It's uncatagorized data.
Many times businesses keep the dark data and just store it, as today it may seem unfeasable to go through it all.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Apple Acquires 'Dark Data' Machine Learning Company Lattice Data
- Thread starter MacRumors
- Start date
- Sort by reaction score
Exactly. But not just about you, about anything. It's uncatagorized data.
Many times businesses keep the dark data and just store it, as today it may seem unfeasable to go through it all.
I thought dark matter was that black blob on my windshield.
Perhaps it can scan, currently random data from millions of RL images, to create yet, unimaginable colour and texture combinations for new watch bands? 
Sorry if I'm being obtuse, but I can't find a video anywhere... What did you watch?
FWIW, I've known plenty of brilliant engineers who are completely incapable of judging the proper level of detail for a given audience... Either they dive deep into the tech, or start talking in car analogies-- but they can't seem to find that middle zone.
"Gibberish" was already taken, plus it doesn't sound as cool. lol.
Jokes aside this sounds really interesting. Adding logic and categorisation capabilities to automation.
Oh yeah, rise of the machines is near!!
The company name is Lattice. Dark Data is a concept.
Based on what little I can find, it looks like these guys specialize data labeling, or maybe unstructured learning (or using one for the other).
When training an AI system, you typically have a "training set" that says "these are all pictures of dogs". When the system studies that set long enough, you can then ask it "does this new picture have a dog in it". The catch is that someone needs to go through all the pictures and find enough with dogs that you can provide as training examples.
This is an incredibly manual, time consuming and error prone process. People are really bad at repetitive mindless tasks. Right now a lot of training sets rely on one of three sources of training data: keyworded data in cloud storage (YouTube, Flickr), Captcha ("prove you're not a robot by finding the pictures of food"), and Amazon Mechanical Turk (farming out to people who get paid 10 cents an image or whatever).
Anything that hasn't run through a system like this, and hasn't been labeled, or at least categorized, is kind of out of the reach of deep learning networks-- it's therefore "dark". Hidden in the shadows.
It sounds like Lattice is trying to further automate the process.
Yes. Much needed and much welcomed changes. And I think those changes are rightfully coming soon, especially being that Apple hired a team of engineers to revamp Siri approximately a year ago.
Apparently Apple want a screen included, so possibly the Echo Show?
Either way, can't wait to see what Apple has been working on come 2022.