Delivering Predictive Outcomes with Superhuman Knowledge
February 13, 2018 Pankaj Goyal, VP, AI Business & Data Center Strategy, Hewlett Packard Enterprise
Massive data growth and advances in acceleration technologies are pushing modern computing capabilities to unprecedented levels and changing the face of entire industries.
Today’s organizations are quickly realizing that the more data they have the more they can learn, and powerful new techniques like artificial intelligence (AI) and deep learning are helping them convert that data into actionable intelligence that can transform nearly every aspect of their business. NVIDIA GPUs and Hewlett Packard Enterprise (HPE) high performance computing (HPC) platforms are accelerating these capabilities and helping organizations arrive at deeper insights, enable dynamic correlation, and deliver predictive outcomes with superhuman knowledge.
As Jim McHugh from NVIDIA recently pointed out, many of today’s companies aren’t drowning in the data deluge, but rather are operating with an insatiable desire for even more data in order to increase their understanding of their business and customers. To make sense of all of this information, companies need tremendous parallel processing power and carefully trained algorithms that are able to detect patterns in large amounts of data, learn independently as more data is introduced to the model, and make predictions based on historical datasets.
Big data and analytics brought us information and insight; AI and deep learning are turning that insight into superhuman knowledge and real-time actionable intelligence. Techniques like deep learning are allowing machines to achieve superhuman levels of perception, enabling companies to leverage predictive compute to automate routine tasks, address problems before they occur, and achieve new levels of competitive advantage.
The human brain is perhaps the world’s most powerful parallel processor, because it can sort and process vast quantities of information, break down complex problems in parallel, and think abstractly in ways that computers still cannot. Because of this, the human brain is still much more efficient than machines in many areas; however, as deep learning algorithms improve, there are some tasks where computers are beginning to surpass human abilities. Here are a few examples:
- Image analysis – Trained deep learning systems can now recognize and classify details in images better and faster than a human can. They can even distinguish significant from unimportant information within images, which can be particularly useful for areas like healthcare, environmental monitoring, and satellite imagery.
- Speech recognition – Deep learning is finally making speech recognition accurate enough to be useful outside of controlled environments. Computer scientist Andrew Ng predicts that as speech recognition capabilities improve to near 99% accuracy, it will become the primary way that humans interact with computers.
- Medical diagnosis – It’s estimated that most people will receive an incorrect or late diagnosis at least once in their lives, sometimes with dire consequences. Deep learning algorithms can pick up on subtle patterns in lab results or charts that doctors can miss, diagnose with higher accuracy, or quickly identify abnormalities in medical images to help doctors pinpoint potential problems faster and more accurately.
NVIDIA GPUs, which allow large computational tasks to be split into thousands of pieces and processed simultaneously, are becoming an essential hardware element for deep learning applications. In fact, it’s estimated that 97% of all new supercomputing nodes now include GPUs. Today, NVIDIA GPUs are acting as the engine of deep learning applications, powering some of the most advanced deep neural networks that offer the speed, accuracy, and scale to drive true AI computing.
HPE offers a new portfolio of purpose-built platforms and services capabilities to help companies simplify the adoption of AI and deep learning. Many organizations lack the expertise, resources, infrastructure, and integration capabilities required to deploy deep learning systems, and HPE’s platforms are the foundation of a high performance compute infrastructure that will be needed to take advantage of deep learning.
Deep learning has sparked a new model of computing, one where machines are reaching superhuman achievements in areas like image analysis, speech recognition, medical care, and much more. Together, HPE and NVIDIA are designing, building, and delivering computing solutions that will allow organizations to harness the full power of deep learning and fully exploit superhuman intelligence to transform their business.
To learn more about how to deliver predictive outcomes with superhuman knowledge, I invite you to follow me on Twitter at @pango. You can also follow @HPE_HPC and @NvidiaAI for the latest HPC and AI news and updates.