All the latest quantum computer articles

See the latest stories on quantum computing from eeNews Europe

Friday, March 24, 2017

Another startup aims for embedded AI

By Nick Flaherty

A spinout from the University of Michigan has raised $9m to develop 'AI-on-a-chip' in a further demonstration of the move of artificial intelligence into the Internet of Things. The move comes just days after ARM and Xilinx announced strategies fo emebdded AI and other startups such as Imagimob and Graphcore are targetting embedded chips.

The company, previously called Isocline, relied on grants and angel investors to develop the underlying technology for its platform and the funding aims to build up the engineering team to develop the chip and software.

At a time when data connections are still unreliable and constrained and even children's toys are getting hacked, it is increasingly important to untether the AI in these devices from the cloud to make them more secure, responsive, seamless, and easily integrated. Because Mythic's local AI platform is untethered and inherently more private, device makers, security platform, and monitoring companies see it as being transformative for their products.

Mythic was born out of the Michigan Integrated Circuits Lab at the University of Michigan in 2012 – a lab known for its world-class low power chip design and multiple venture-backed spin-outs. Co-founders Mike Henry and Dave Fick developed a new deep learning deployment (or inference) model – based on hybrid digital/analog computation– that eliminates costly processors and transfers the deep learning computations to the memory structures storing the algorithm parameters – all while extending battery life by 50x. This design allows the company to essentially put desktop GPU compute capabilities onto a module the size of a shirt button and data-center compute capabilities onto a card-deck sized platform– offering a new level of trust and security to the growing number of smart devices on the market today.

"Other local AI solutions compromise on what consumers and commercial companies care about – battery life, throughput, and accuracy," said Michael Henry, co-founder and CEO of Mythic. "We saw early on the opportunities afforded AI from processor-in-memory technology and new methods of computing outside of the binary 1 and 0 world. This would become a foundation for the Mythic platform that revolutionizes local AI design and performance."

Markets that Mythic will be targeting initially include smart home, action cameras, sophisticated healthcare systems, security and monitoring for commercial and home use and drones for industrial applications. Down the road, Mythic sees interesting use cases for robotics, autonomous vehicles, AR, and VR.

"Mythic is pushing the performance boundaries of local AI so any device can become a true universal assistant," said Steve Jurvetson, Partner at DFJ. "The way local AI is being engineered today is equivalent to trying to make a propeller plane fly faster by stripping down cargo, or adding 18 propellers – working with technology that stalled just doesn't make sense. Just as jet design put flight on a new trajectory, Mythic is doing the same for local AI."

The company is forming engagements with early-adopter customers to field test the technology, with volume shipments projected for mid-2018. The funding round was led by DFJ and includes Lux Capital, Data Collective, and AME Cloud Ventures. 

No comments: