Nvidia and DFKI hope to build an AI Model Store for industrial users

Nvidia and DFKI, a German AI research hub, hope to build an AI Model store so companies can shop for practical apps, according to Andreas Dengel. (Nvidia)

Nvidia has become a shareholder in DFKI, Germany’s top business research institution that investigates software based on artificial intelligence.

Nvidia and DFKI, also known as the German Center for Artificial Intelligence, already have been cooperating on AI research. The new shareholder status “opens up new perspectives for application-oriented neural network research,” said Andreas Dengel, scientific director for DFKI’s research department on smart data and knowledge services, in a statement issued Wednesday.

The partnership will be used to make complex AI algorithms available for practical industrial use.  Dengel envisions an AI Model Store where a variety of machine learning methods are offered.  There will be a fully automated architecture search where companies can select the architecture that fits their needs, then can apply data to the pre-trained approaches and receive feedback.

Free Daily Newsletter

Interesting read? Subscribe to FierceElectronics!

The electronics industry remains in flux as constant innovation fuels market trends. FierceElectronics subscribers rely on our suite of newsletters as their must-read source for the latest news, developments and predictions impacting their world. Sign up today to get electronics news and updates delivered to your inbox and read on the go.

Nvidia invented the GPU used in many high performance computers that are essential to AI work.  The company was the ninth largest semiconductor maker globally in 2019.  Nvidia named DFKI a GPU research center in 2016 for its research in analyzing multimedia data and in 2018 Nvidia conferred its Pioneer Award on DFKI for its work on neural network processing.

DFKI runs several Nvidia DGX systems in a Mellanox InfiniBand network in Kaiserslautern, Germany. The network includes the first DGX-2 supercomputer in Europe.  With that network and fast data processing, high-resolution satellite images can be analyzed quickly to help coordinate disaster relief work, among other practical uses.  In all, DFKI computers using DGX provide 17 petaflops of computing power.

In a blog, Nvidia said an unnamed company is already testing a prototype AI sandbox and catalog that DFKI built to let users try deep learning.  The catalog provides 35 top neural networks for imaging. A commercial use is planned for next year.

Europe is trying to beef up AI research, and in mid-March the European Commission granted $54 million to a new AI research alliance using the acronym of CLAIRE (Confederation of Laboratories for Artificial Intelligence Research in Europe)

DFKI has about 250 AI projects across 20 departments. One team is helping Germany’s federal bank apply AI, while another conducts 600-hour tests of car engines to make predictions with AI.  

Nvidia and many large semiconductor makers are pursuing AI research on multiple fronts.  Nvidia announced Monday it was joining the COVID-19 High Performance Computing Consortium, a group founded by IBM with the U.S. government, to provide free access to 30 supercomputers to further scientific research into COVID-19 treatments and vaccines.

RELATED: COVID-19 prompts spike in liquor sales, computing research

Suggested Articles

Company’s preliminary third quarter shows better than expected market, with gain over earlier prediction

Low-level autonomy vehicles will grow by 11% a year and there won’t be any produced at the highest autonomy before 2024, according to IDC.

Nvidia’s GPU Cloud hub will be integrated into vSphere, Cloud Foundation and Tanzu