How Facebook Can Help Make Your Next Car Safer

To anyone who knows a teenage driver, you can’t help but worry about the inevitable near-miss (or worse) accidents caused by a distracted driver checking their Facebook page. However, your next car just might be a whole lot safer because of Deep Neural Networks, or DNN for short, technology being developed at Facebook and scores of other companies up and down Silicon Valley. You need look no farther than Yann LeCun’s Facebook page, but not while driving please, to see what Facebook’s Director of Artificial Intelligence is up to with DNNs. Besides his job at Facebook, Yann also is a professor at NYU’s Computer Science department where he helped pioneer many current advances in the field. But there is more to the Facebook-NYU connection than your typical Silicon Valley university relationship, it stems from the core of the innovation driving DNNs as one of today’s leading Machine Learning approaches, the massive amounts of big data used to train DNNs.

DNN algorithms are not particularly new. What is relatively new is the use of DNNs combined with the massive amounts of unstructured big data including voice, images, and video stored by today’s top social networking and search sites combined with unparalleled levels of performance provided by GPUs to crunch all of that data through DNNs in a cost and power efficient manner. One of the key mathematical algorithms used in DNNs is the Fast Fourier Transform, or FFT. GPUs are particularly well suited to processing FFTs. For DNNs, Facebook recently made this even more true when LeCun and his collaborators released the new fbFFT library.

If the fbFFT paper is a bit too technical for you, Jeremy Howard’s recent Ted Talk on Machine Learning helps explain the technology in simpler ways through lots of examples. As the founder of Silicon Valley startup Enlitic, Jeremy knows a thing or two about machine learning.

Now Facebook may or may not have any interest in self-driving cars, but the same DNN technology that can automatically identify your friend’s picture on a Facebook page is much the same as the technology that can already help an automobile identify pedestrians in a crosswalk or slowing traffic ahead. This week at the CES Consumer Electronics Show, NVIDIA introduced a host of new technologies including the new Tegra X1 mobile super chip, capable of processing over 1 TeraFlop a second of DNN instructions to the new NVIDIA Drive PX auto-pilot car computer which will make it easier than ever for automotive manufacturers to integrate advanced DNN technology into future vehicles.

While you can’t yet buy a car with the Drive PX auto-pilot computer, developers today can start writing software for it on any NVIDIA GPU platform, from the $192 Jetson TK1 developer kit to the GeForce GTX 980, the world’s most advanced GPU utilizing the same Maxwell technology used in the upcoming Tegra X1.

But for now, Facebook in cars should remain for passenger user only. For more info on the Tegra X1, Drive PX, and other new NVIDIA technologies watch our CES press conference below.



Broadcast live streaming video on Ustream

Advertisements

About Marc Hamilton

Marc Hamilton – Vice President, Solutions Architecture and Engineering, NVIDIA. At NVIDIA, the Visual Computing Company, Marc leads the worldwide Solutions Architecture and Engineering team, responsible for working with NVIDIA’s customers and partners to deliver the world’s best end to end solutions for professional visualization and design, high performance computing, and big data analytics. Prior to NVIDIA, Marc worked in the Hyperscale Business Unit within HP’s Enterprise Group where he led the HPC team for the Americas region. Marc spent 16 years at Sun Microsystems in HPC and other sales and marketing executive management roles. Marc also worked at TRW developing HPC applications for the US aerospace and defense industry. He has published a number of technical articles and is the author of the book, “Software Development, Building Reliable Systems”. Marc holds a BS degree in Math and Computer Science from UCLA, an MS degree in Electrical Engineering from USC, and is a graduate of the UCLA Executive Management program.
This entry was posted in Uncategorized. Bookmark the permalink.