The Stack Archive

Facebook donating 25 state-of-the-art GPU servers to AI research in Europe

Thu 25 Feb 2016

The artificial intelligence division of Facebook is to donate 25 powerful GPU-based servers to research institutions in the European Union with a view to accelerating progress in the science of artificial intelligence.

According to the company’s post today, the first donation will go to Prof. Klaus-Robert Müller of The Technical University of Berlin, currently occupied with machine learning at the university’s Institute of Software Engineering and Theoretical Computer Science. Müller’s immediate scope with the new equipment, which will also come with technical support, will be research into image analysis of breast cancer and the chemical modelling of molecules.

Dr Müller commented in the announcement “This partnership comes at exactly the right moment for AI research in Germany. It will help us as we study two very hard and computationally intense learning problems around image analysis of breast cancer and chemical modelling of molecules. The new servers will help us speed up our research cycles and do better AI research much faster.”

The initiative comes through the Facebook AI Research (FAIR) program, which was set up in 2013 and handles a wide variety of interesting projects, most particularly in the subject perhaps dearest to Facebook’s corporate heart – image recognition.

In addition to the provision of the GPU servers – technical details of which have not been revealed yet – Facebook will work with the EU recipients to ensure that the partners have suitable software to run the servers, and will envoy researchers to participate in the new server-based projects.

Facebook is opening the program up to the broader European community, inviting research institutions to make application proposals for GPU donations by submitting a one-page proposal to gpuapplication@fb.com . The criteria for applicants can be found here, and includes the stipulation that participants must publish data sets resultant from their GPU research, and that the institutions – which must be publicly funded, academic or governmental in nature – must already have a laboratory or research group working on problems of artificial intelligence. Institutions with a reputation for collaboration within the wider scientific community will receive further preference.

Send us a correction about this article Send us a news tip

Do NOT follow this link or you will be banned from the site!