Monday , 26 June 2017
Home >> A >> Apple >> Apple acquires AI association Lattice Data, a dilettante in unstructured ‘dark data’, for $200M

Apple acquires AI association Lattice Data, a dilettante in unstructured ‘dark data’, for $200M


As vast tech companies rigging adult to make a stronger pull into appurtenance training and synthetic intelligence, Apple has acquired a association to fill out a possess capabilities in the area.

Specifically, Apple has picked adult Lattice Data, a association that relates an AI-enabled deduction engine to take unstructured, “dark” information and turn it into structured (and some-more usable) information. We’ve listened from a singular source that Apple has paid a cost of around $200 million.

The understanding was sealed a integrate of weeks ago, a source said, and about 20 engineers have assimilated a incomparable company.

A source initial alerted us to this, and we’ve perceived a customary Apple confirmation. “Apple buys smaller record companies from time to time and we generally do not plead a purpose or plans,” an Apple orator told TechCrunch.

Lattice quietly lifted during slightest $20 million appropriation from GV, Madrona and InQTel before exiting from secrecy final year. Lattice was founded in 2015 and has mostly remained underneath a radar, though it has a notable pedigree.

The association was co-founded by Christopher Ré, Michael Cafarella, Raphael Hoffmann and Feng Niu as a commercialization of DeepDive, a system created during Stanford to “extract value from dim data.”

Ré, a highbrow during Stanford, won a MacArthur Genius Grant for his work on DeepDive and is now arch scientist during Lattice. Cafarella, who started as Lattice’s CEO though is now a CTO, is a highbrow during a University of Michigan and is famous as one of a co-creators of Hadoop. Niu is Lattice’s arch engineering officer. Cafarella and Hoffmann (who has given changed to Google, according to his LinkedIn) also worked on developing DeepDive.


The CEO of Lattice is Andy Jacques, a seasoned craving executive who assimilated final year.

What accurately is dim data? Our connected, digital universe is producing information during an accelerated pace: there were 4.4 zettabytes of information in 2013, and that’s projected to grow to 44 zettabytes by 2020, and IBM estimates that 90 percent of a information in existence currently was constructed in a final dual years.

But between 70 and 80 percent of that information is unstructured — that is, “dark” — and therefore mostly obsolete when it comes to estimate and analytics. Lattice uses appurtenance training to radically put that information into sequence and to make it some-more usable.

Think of it in terms of a variety of information but labels, categorization or a clarity of context — but with a certain implicit value that could be unbarred with correct organization.

The applications of a complement are manifold: they can be used in general policing and crime solving, such as this work in perplexing to uncover human trafficking; in medical research; and to assistance classify and parse paleontological research. It also could be used to assistance sight AI systems by formulating some-more useful information feeds.

It’s misleading who Lattice has been operative with, or how Apple intends to use a technology. Our theory is that there is an AI play here: Our source pronounced that Lattice had been “talking to other tech companies about enhancing their AI assistants,” including Amazon’s Alexa and Samsung’s Bixby, and had recently spent time in South Korea.