Nvidia builds connections to AI researchers

June 27, 2016 // By Christoph Hammerschmidt
Deep Learning is rapidly gaining importance in the development of automated vehicles. To expand its position and to secure access to the latest developments in this area, chipmaker Nvidia has entered a closer relationship with the German Artificial Research Center (DFKI).

Nvidia appointed the Kaiserslautern, Germany, based institution a GPU Research Center. In Nvidia’s parlance, GPU Research Centers are institutions who apply Graphics Processing Units (GPUs) in specific research areas. The GPU manufacturer is rather picky with this attribute; only leading scientific institutes receive Nvidia’s appointment.


The DFKI has been selected for its outstanding research activities in the areas of text, image, video and social media analysis with GPU hardware as the underlying computing technology. In particular, the Sentibank visual sentiment analysis framework contributed to Nvidia’s move. Sentibank has been further developed by DFKI. Being the first German Deep Learning research institute admitted to the elitist circle of GPU Research Centers, DFKI scientists now have preferred access to hardware and software under development from Nvidia as well as to scientific exchange platforms and networks.


GPU hardware provided to the DFKI will be applied in the first place in the area of Deep Learning. Currently, many scientific breakthroughs in the realm of Artificial Intelligence are associated to Deep Learning. In particular, Deep Learning is a hot candidate for the technologies that will enable self-driving vehicles to orientate themselves even in environments that are subject to dynamic changes.


One of the reasons why Deep Learning is currently so successful is the availability of high-performance GPUs that can run highly parallel training algorithms for Deep Learning. “Originally developed for graphics processing, Nvidia GPUs have made decisive contributions to the success of Deep Learning techniques”, said Damian Borth who oversees the Deep Learning competency center at the DFKI.


In the past, DFKI researchers already performed projects related to automotive applications. While for DFKI’s appointment to a GPU Research Center, automotive research activities did not play a direct role, it is “well possible” that in the future, automotive topics could become a hotspot of the collaboration, Borth explained. Over the past couple of years, Nvidia has successfully worked to get a foothold