skip to main content
:::
:::

Special Report

A New Era of Computing Powered by GPUs and AI

In the keynote speech, Dr. Deepu Talla, Vice President & General Manager of Autonomous Machines, NVIDIA Corporation, stressed that artificial intelligence in the form of deep learning is transforming, and that GPU computing has emerged as the standard of accelerated computing by increasing performance at Moore’s law squared. He further illustrated how NVIDIA’s work in high-performance computing and AI influences autonomous vehicles, robotics and intelligent video analytics.

Dr. Deepu Talla, Vice President & General Manager of Autonomous Machines, NVIDIA Corporation.

Dr. Deepu Talla, Vice President & General Manager of Autonomous Machines, NVIDIA Corporation.

The progress of GPU computing is one of the reasons why AI and deep learning can rapidly advance and be deployed in more areas. Dr. Talla listed the three chapters in AI: the first chapter is about training and creating neural networks, followed by the second which is inference, or deploying the neural networks. Dr. Talla pointed out that in fact we are all using some form of AI inference. For instance, when we are doing a Google search, the request from our mobile device is sent to the cloud where the neural network runs.

The third chapter, which is also the chapter he is the most excited about, is about bringing AI from the cloud to industries, machines, and devices. This new development brings forth a new generation of machines—the autonomous machines, which will level the playing field for industries. The trend will also offer a lot more business opportunities, as these new machines and devices require the collaboration of various industries and professionals, such as sensor industries and engineers and programmers.

Dr. Talla then provided three examples of AI applications: autonomous vehicles, robotics, and video analytics for smart cities. He indicated that safety is the first and foremost thing to consider when developing a self-driving car, and this includes the ability to not only protect the driver and passengers inside the vehicle, but also to detect pedestrians and other obstacles in the road. NVIDIA’s platform provides a safe space for engineers and manufacturers to train the vehicle’s neural networks and run them through scenario simulations so that the autonomous vehicles can drive safely on actual streets.

Robotics is more sophisticated than autonomous vehicles, as Dr. Talla explained, because autonomous vehicles only need to go from point A to B without touching anything, but robots have to perform more complicated tasks. In addition, industrial robots usually need to work in close proximity with humans, therefore proper training before deployment is extremely important. NVIDIA offers a simulation platform and toolkit for developers to safely train their robots in a virtual environment.

The third application of AI is in active video analytics to build smart cities. Technology that is capable of processing and analyzing large amounts of camera footage is needed to provide insight for policy making and other functions.

“The third chapter of AI is fresh. All of us have a chance to write it in the next 10 to 20 years,” concluded Dr. Talla. “It’s all about bringing AI and deep learning into industries.”

TOP