Verne Global

AI / ML / DL | HPC | Tech Trends |

19 September 2018

Trends Advancing Industrial HPC

Written by Brendan McGinty (Guest)

Brendan McGinty is Director of Industry for the National Center for Supercomputing Applications (NCSA), University of Illinois at Urbana-Champaign.

The global research community, be it academic or applied, has a rich heritage of driving innovation and breakthroughs enabled by HPC. From a business perspective, what are the industry trends around supercomputing and HPC in advancing innovation in key markets?

Three particularly hot topics

In the past year, three terms have been all the buzz in supercomputing: AI, machine learning, and deep learning. Supercomputing conferences feature hardware and software vendors alike offering their latest and greatest to provide solutions in these three areas. Universities, including supercomputing centers, have more academic professionals and students involved in these areas than ever. From a research perspective, including from an applied standpoint, there is significant time and resource being put into advancing capabilities in AI, machine learning, and deep learning.

Perspective on how to provide solutions

I’m a longtime corporate consultant. As such, I learned long ago that if I didn’t understand what companies needed — or thought they needed — my engagements were going to be of the one-and-done variety. I now have the best job in the world because I get to consistently work with and around people smarter than I am. Our team at the National Center for Supercomputing Applications (NCSA), the largest industry HPC outreach in the world, is comprised of mostly PhDs that are experts in HPC and a domain.

As such, we consult with and support companies all around the world with grand challenges in several domains, including data analytics (including AI, machine learning, and deep learning as well as geospatial and other analytics methodologies), bioinformatics (think genomics), modeling and simulation, visualisation (particularly applied for industry), and cyber infrastructure (compute and security.) Talk about covering some ground — we have to keep up with the latest advancements from an application and compute standpoint constantly so that we can best support our corporate partners. We always have to be consultative — hearing corporate needs and challenges and providing ideas and solutions that benefit their business operations.

Corporate perspective

The companies I talk to, which are many and around the globe, have similar needs and express those needs in similar ways. I oversimplify their statements accordingly:

  • We have a lot of data and we don’t know what we can do with it
  • We need machine learning
  • Integrating AI will make us better
  • We need to ensure that our data is secure.

Nearly all of the companies talk of wanting to be disruptive — innovative — to separate from the competition, stay a step ahead, and provide next-generation solutions to tomorrow’s challenges. They think they need deep learning but can’t tell us why. They’re pretty sure that AI could help them because they heard it somewhere but they don’t know how that gets applied to their challenges. How do we bridge those gaps and help to explain why these solutions can help and how to make that happen?

How to address

There is a lot of noise out there. How do you weed through it to get to what is beneficial to a particular business?

While we have expertise in many domains including data analytics, that domain alone includes the big three – AI, machine learning, and deep learning — and others, like geospatial. To address those why and how questions, we have to help educate on what’s possible.

While white papers and lengthy presentations can help to inform, industry is different. Time and attention is limited. Seemingly hundreds of things are being juggled simultaneously. While unfortunate, reality is that not many corporate executives or business decision-makers are taking or have the time to read white papers. There are two areas that I have found that are effective — that stick and help businesses to, in short order, realise the benefits of emerging technological solutions like AI, machine learning, and deep learning:

  1. Demonstration of real-world use – a visualisation, video, or animation, perhaps combined with audio, of an actual project or accomplishment, that can paint the picture in very short time of the benefit of a particular solution set.
  2. An ideation meeting — some call it discovery, analysis, initiation, the term is less important than the function. In these meetings, a wider range of (potentially) interested parties — corporate leaders and participants on one side and solution providers like domain experts on the other — come together to discuss current needs, future plans, and even dreams. They discuss what’s possible or what might be possible, all together, to enhance communication between themselves and internally in each group as well. Come out with an action plan, both short-term and longer, and you have success.

Summary

The three hot topics these days in HPC are AI, machine learning, and deep learning. That’s the easy part — knowing that — because everyone is talking about those areas. The trick — the challenge — is how to make it all mean something. That something involves addressing real corporate challenges with solutions in these three areas that are truly applied, pertinent to the business, and resulting in enhanced business operations.

Editor's Note: In the run-up to SC18, Verne Global is hosting an AI and HPC Field Trip to Iceland between the 23-25 October, and a further AI and HPC Field Trip (due to overwhelming demand) between the 26-28 February 2019. If you are interested in attending please contact Bob Fletcher: bob.fletcher@verneglobal.com

Share:FacebookTwitterLinkedInmail

Sign up for the Verne Global newsletter

Opinion, thought leadership and news delivered directly to your inbox once a month.

SIGN ME UP