As a Fortune 250 company with over 220 locations worldwide, Qualcomm are undisputed giants of the semiconductor space.
But what part does artificial intelligence have to play in a company that is already designing and manufacturing cutting-edge technology?
To find out, AI Business caught up with Gary Brotman, Qualcomm Technologies’s Director of Product Management, who told us about Qualcomm’s broad approach to AI as well as its impact on their latest products.

 

Brotman_Gary_2016_headshot_usethisone cropped 1

Gary Brotman of Qualcomm Technologies

Having emerged as market leaders in mobile technology with their Snapdragon processors, the first of which was released back in 2007, Gary tells us Qualcomm is now investing their efforts more heavily into machine learning – or ‘machine intelligence’ as they call it.

“Machine learning is an umbrella category – machine intelligence is the way we frame it. We are approaching this holistically to apply the right technology for on-device machine intelligence versus focusing on one or two of machine learning executions..

 

Gary explains that Qualcomm’s focus is directed by use cases, and the demand they see in the marketplace and from their customers. “We’ll try the appropriate algorithm and software to solve the problem, all with the goal in mind of being able to harness the compute capabilities of our core Snapdragon processor, our SoCs [System-on Chips]”, he explained.

 

Over the past 12 to 18 months the efforts of the R&D and commercial teams at Qualcomm have been software-specific, Gary says:

“Last year we released a product called Snapdragon Scene Detect, which, as you can guess by its name, is a scene detection SDK to add deep learning capabilities to gallery and live camera apps for a handset, enabling users to identify images and allowing them to categorise images by their type. This technology is not new, but we brought it to the handset with Snapdragon acceleration for the first time”.

Another mobile product released last year, also with specific benefits to the end user and enterprises, is Snapdragon’s Smart Protect technology, which utilizes machine learning to provide real-time, on-device behavioural analysis to identify applications containing malware. “Unlike signature-based anti-virus software, Smart Protect is demonstrably more successful at seeking out and preventing malware from harming devices – making Android products, for instance, as rock solid as they can be”.

But there have been more recent major developments; just six weeks ago, Qualcomm announced their Snapdragon Neural Processing Engine SDK, powered by the Qualcomm Zeroth Machine Intelligence Platform. According to Qualcomm it is ‘optimized to utilize Snapdragon’s heterogeneous compute capabilities to provide OEMs a powerful, energy efficient platform for delivering intuitive and engaging deep learning-driven experiences on device’.

NPE Graphic Qualcomm

Gary explains that the platform is data- and platform-agnostic: “the whole effort is on-device edge compute, and what was 12 months ago or even today sporadic behaviour with respect to deep neural network workloads running on an edge device, has now become more commonplace to the point where you will have multiple networks for multiple applications running simultaneously”.

In terms of its impact on the enterprise, Snapdragon Neural Processing Engine SDK is said to support companies in a broad range of industries, including healthcare, automotive, security and imaging, to run their own proprietary trained neural network models on portable devices.

Gary tells us that Qualcomm will see initial early access partners implement the platform over the next few months, but by the end of the year it will be more broadly available.

 

And the goal going forward? “We are focusing on making Snapdragon the ideal mobile platform for running neural networks or machine intelligence software experiences at the edge today and into the future – the research effort is very active and ongoing, and we should be able to share more news in the near future”.

 

Artificial intelligence has clearly become an integral component of Qualcomm’s strategy as they implement their latest products into the enterprise, with the potential to redefine their approach to to IoT, mobile, automotive, drones and cameras.

As the enterprise as a whole takes steps forward in becoming AI-enabled, AI Business is committed to providing up-to-the-minute news and conversations on the practical application of artificial intelligence in the business world.

For the latest, follow us on Twitter @Business_AI and join us on LinkedIn – AI Business Community

Feature image credit: Flickr