IEEE Spectrum June, 2016 - 19

reSoUrceS_StArtUPS

nervAnA systeMs
turning neural networks
into a serviCe

eep learning is Silicon Valley's
latest and greatest attempt at
training artificial intelligence to
understand the world by using data to train
algorithmic models with many layers of processing. One promising approach to building
such layers is to loosely imitate how neurons in the brain work. And a startup called
Nervana Systems, based in San Diego, aims
to make such neural-network-based deeplearning AI even more widely available by
turning it into a cloud service.
The customer is any industry with big-data
problems to solve. "Our goal is to really be the
world's platform on which you do artificial intelligence," says Naveen Rao, cofounder and
CEO of Nervana.
Rao says Nervana has a different philosophy than many other deep-learning
startups, which are "trying to compete
head-to-head with the Googles of the
world." He thinks it will be tough for even the
scrappiest startups to beat out Google's
huge data-science team in tackling the
toughest computational problems. Instead,
Nervana wants to develop and sell deep-

getty images

learning AI as a service for the many companies that have more mundane big-data
problems but lack the in-house expertise
and systems to handle them.
"From the start, we've been focused on
making deep learning fast, easy to use, and
scaling," Rao says.
On 29 February, Nervana officially debuted its Nervana Cloud service. Nervana
Cloud is a hosted hardware and software
platform that allows any organization to develop its own deep-learning solutions tailored to the specific problems of its industry,
be it health care, agriculture, finance, energy,
or something else.
Nervana's deep-learning AI has already been at work for several companies.
Blue River Technology is a precisionagriculture company that uses computer
vision and robots to improve farming efficiency by removing unwanted plants and
making decisions based on the condition
of individual crop plants. By using Nervana's service, Blue River improved the
reliability of its robots' ability to identify individual plants.

Another company called Paradigm, in
Houston, used Nervana Cloud to better detect underground features within scans of
3-D volumes that could indicate good locations for oil drilling. The improved accuracy translated into more efficient drilling
decisions that reduced the time and money wasted on locations that might not yield
worthwhile oil deposits.
Nervana's deep-learning software currently runs on Nvidia graphic chips because
such chips have proved surprisingly useful to
many problems for which numbers must be
crunched in parallel. But the startup is developing its own optimized hardware.
Rao says Nervana's new architecture for
deep learning may involve multiple specialized chips that could work together in concert. That would eventually enable much
larger versions of the brain-inspired neural networks that form the foundation of
deep-learning AI, he says. Such larger neural networks could enable deep-learning AI
to sift through more challenging big-data
problems involving high-resolution images and video.
Founded in 2014, Nervana has about 40
employees split between the original office
in San Diego and a somewhat larger branch
in Palo Alto, Calif. That makes it a fairly lean
operation in a deep-learning field filled with
tech giants such as Google and IBM. Still,
the startup has already raised more than
US $24 million in seed funding from venture capital firms such as Allen & Co., AME
Cloud Ventures, CME Group, DCVC, DFJ,
Fuel Capital, Lux Capital, Omidyar Network,
and Playground Global.
There's a lot of low-hanging fruit for deeplearning AI to pluck in terms of the more common big-data problems facing many different
industries, says Rao. But to thrive, Nervana
still needs to figure out how to sell its deeplearning cloud platform as a valued service to
more companies.
"We're doing deep-learning [software solutions as a complete package]; what will
people pay for that?" Rao says. -JereMy hSu
Location: San Diego and Palo Alto, Calif.
Founded: 2014 Employees: 40 Funding:
Over $24 million

SPectrUm.ieee.orG

North AmericAN

JUN 2016

http://SPectrUm.ieee.orG

Table of Contents for the Digital Edition of IEEE Spectrum June, 2016