Ex-NASA chief's neural computing startup reveals 256-core processor and voice recognition system

[midian182]

Daniel Golding, the former National Aeronautics and Space Administration (NASA) director, has just officially unveiled his new company, KnuEdge, which has been in stealth for the last ten years and already raised $100 million in funding.

KnuEdge, which was formerly called Intellisis, has spent the last decade working on two products that focus on neural computing: KnuVerse, a military-grade speech recognition and authentication engine, and KnuPath, a processor designed to run machine learning algorithms.

The company says that KnuVerse was produced in “mission-critical battlefield conditions.” It can recognize and authenticate the voice of an individual even in the noisiest of environments. Away from military applications, KnuEdge says the technology can be used in consumer products where background interference causes problems with speech recognition, and its ability to confirm a person’s identity after just a few words will no doubt be of interest to many large companies.

"Think of it as a version of Google Now on steroids that has been proven by the U.S. government and military," said Jim McGregor, principal analyst at Tirias Research.

The company’s other product, the KnuPath processor, came about after it discovered that traditional CPUs and GPUs weren’t up to the job when it came to creating the KnuVerse speech recognition technology. To get past these performance issues, a new team was assigned with creating an application-specific integrated circuit (ASIC).

The low-wattage chip features 256 cores and 16 bidirectional I/O paths that provide 320 Gbps of throughput. KnuEdge says the processor can scale to over half a million devices, and offers inter-rack latency of around 400 milliseconds.

KnuEdge plans to make its APIs available worldwide as freemium service, with the KnuPath chips set for mainstream release in the second half of the year. In addition to the $100 million in funding, the company has also generated $20 million in revenue – not too much of a surprise, considering the number of organizations that are investing in AI and machine learning technology.

Permalink to story.

https://www.techspot.com/news/65125-ex-nasa-chief-neural-computing-startup-reveals-256.html

 

[davimill]

The man's name is spelled Goldin, not Golding. Before he became the NASA Administrator, he was the manager of a division at TRW while I was working there in the 1990s on the Minuteman missile program. Ref: https://en.wikipedia.org/wiki/Daniel_Goldin

 

[Stryder007]

So this fellow gets to take the tech he developed while being paid by tax payer dollars then turn around and open a company to sell said tech for personal gain,wake up people your being used

 

[Bruce Fraser]

So this fellow gets to take the tech he developed while being paid by tax payer

What is your source for that claim? Please, something more substantial than "Duh, everyone knows that."

 

[Tosikko]

The approach seems kind of aggressive, but I'm liking the results they achieved. I'd expect the competition to catch up pretty quickly though.

So this fellow gets to take the tech he developed while being paid by tax payer dollars then turn around and open a company to sell said tech for personal gain,wake up people your being used

That kind of logic was born dead. Besides, the guy seems to have left NASA already back in 2001.

 

[Yynxs]

So this fellow gets to take the tech he developed while being paid by tax payer dollars then turn around and open a company to sell said tech for personal gain,wake up people your being used

-----------------------------
All tech developed by the government is actually 'public'. Should you have the capability you can request the specifications on the chips and algorithms be released to you (with copying fees etc) and you can start your own company.

This man is just taking the path of politicians and ex-government officials into the private sector albeit, his own startup. What should be being asked is, "who exactly are the 'investors' in this startup?

From my privacy paranoid point of view, I'm seeing real time voice tracking of cellphone conversations, restaurant or public area conversations, and possibly, the capability (coupled with that fractional foot device tracking from an earlier TS article) of putting advertising based on what two or more people are saying, around them as they move. That's aside from the real gem this is for police and intel agencies who no longer have to rely on line-of-sight or lip reading to "hear" what the 'bad guys' are saying.