Soccer team in a huddle

Accelerating Enterprise Intelligence: SAP Announces NVIDIA Partnership

Feature Article | May 10, 2017 by Juergen Mueller

In a complex and competitive world, the organisms that survive are the ones that are agile. This means that they sense the environment, quickly analyze a situation, and then take the right action based on that analysis.

To maximize value for your customers, these three capabilities are increasingly critical to your company, and ours.

To make this possible, SAP has strengthened its ability to build intelligent solutions through expanding our strategic partnership with NVIDIA, as just announced at NVIDIA’s GPU Technology Conference (GTC).

Fast Training of Machine Learning Models

I’ve written previously about the power of artificial intelligence (AI) for the intelligent enterprise and how machine learning will benefit businesses in numerous ways. Being able to do sophisticated analyses of huge data sets to find subtle and powerful data patterns is an integral part of that and an essential element of our machine learning solutions and services. This is where NVIDIA comes in.

SAP is expanding its strategic partnership with NVIDIA

I’ve been very impressed by the speed of NVIDIA’s computing platform, whose Graphics Processing Unit (GPU)-based parallelism powers our training of massive data sets and deep learning algorithms that are the basis for our machine learning applications. With 76 percent of the world’s transaction revenue touching SAP systems, that’s a lot of data. This data is the fuel to train advanced machine learning solutions for the enterprise. A deep learning run that would have taken years in the past now finishes in just a few days.

Imagine this: a single NVIDIA GPU runs over 10 teraflops — that’s a million million operations per second.  This technology is already being used for the most important problems on the planet, such as detecting (and deflecting!) rogue asteroids inbound towards earth and analyzing data from the Large Hadron Collider. Now, we’re applying it to enterprise applications.

SAP Brand Impact: A Computer Vision Application that Demonstrates the Power of NVIDIA Technology

Fast processing time also translates into the ability to learn from huge data sets. For instance, at GTC, SAP showcases SAP Brand Impact, which analyzes a company’s brand exposure. The application was shown thousands of images and videos from social media and other sources. It learned how to recognize logos and other brand information from the images, without the need to explicitly program the software. Given the huge amount of data, and complex processing requirements, this task would have been impossible without powerful technology.

Now that the SAP Brand Impact model has been learned, it can very rapidly scan thousands of new videos and images, so that companies can gain accurate and timely insights into the return on advertising and sponsorship investment, and ultimately their true brand value. Marketers can now take immediate action: For example, if they detected that their brand visibility during a soccer match is lower than what was contractually agreed, they could instantly request more air time for the second half. This is live business.

And SAP Brand Impact is just one application in the first wave of machine learning solutions that are trained and operated on the NVIDIA deep learning platform: SAP Service Ticketing and SAP Accounts Payable are also showcased at GTC. With accelerated responses and reduced customer service costs, SAP Service Ticketing will bring customer satisfaction to the next level, while SAP Accounts Payable is an important step towards advanced finance operations and self-driving accounts payable processing. Indeed, our entire finance product line will ultimately be made intelligent in only a few years.

Integrating Intelligence and Agility into the Fabric of Your Organization

SAP is committed to help our customers become an “agile organism” across the board. You can now constantly sense your environment (with data), then analyze up to thousands of variables at once (with machine learning), leading to intelligent actions (embedded into applications). With NVIDIA, we speed this loop to be several magnitudes faster than would otherwise be possible.

An important priority for us is to keep this sophisticated technology usable and accessible. So, here’s our approach:

  • Our open ecosystem — illustrated by the NVIDIA partnership — shows that we are agile ourselves — constantly able to adapt to the latest technologies provided by our partners.
  • As a user, you’ll find more and more applications enabled by machine learning – intelligent services embedded in SAP Cloud Platform. We are re-imagining all business processes with digital intelligence, aiming to turn all repetitive transactional activities on digital data into software. First proof points include the offerings SAP Cash Application, SAP Brand Impact, and SAP Service Ticketing.
  • As a developer, you can make use of two levels of application programming interfaces (APIs) to add intelligence to your solutions – soon easily consumable via the SAP API Business Hub:
    • Business service APIs, such as those that match invoices to payments
    • Functional level APIs, which perform lower-level tasks like classifying images, recognizing the key concepts in natural language text, or projecting forward from a time series.

Bottom line, machine learning is one of the most important and comprehensive initiatives I’ve been involved in at SAP to date. And with strong partners like NVIDIA at our side, the possibilities are limitless. New applications, unprecedented levels of performance in existing applications, and easy access to machine learning services will allow you to make your own organization agile and intelligent, in a fast and simple manner.

Juegen Mueller is chief innovation officer at SAP.

Tags: , , , , , , , , , ,

Leave a Reply