SURFsara and Intel to collaborate on deep learning

New Intel® Parallel Computing Center

02 MAR 2017

Intel and SURFsara will work together to harness the power of Intel’s new CPU architectures for deep learning. This will enable SURFsara to support researchers to train and develop more accurate and powerful models with neural networks that are more complex and on data sets that are larger than the ones typically trained today.

Deep learning applications

SURFsara and Intel have agreed to collaborate under the framework of the Intel® Parallel Computing Center (Intel® PCC) concept. Within this collaboration, both Intel and SURFsara plan to work on deep learning applications, using high-performance computing. The field of deep learning is of high interest in numerous scientific domains, including the domains of smart energy and smart health. Both for SURFsara end users and for Intel customers, optimized applications on highly scalable parallel and distributed systems will have the ability to advance scientific research using deep learning concepts.

Global scientific community

Combining the expertise available at SURFsara and the advanced Intel® Xeon Phi™ processor architecture, SURFsara and Intel expect to progress on the performance and scalability of deep learning applications. Results of this collaboration will be fed back into  development of future hardware solutions and into applications and algorithms available to the global scientific community.

Artificial neural networks

Prof. dr. Anwar Osseyran, Member Executive Board SURF and CEO of SURFsara: “SURFsara’s aim is to provide open source deep learning tools that will benefit researchers and supercomputing experts not only in The Netherlands but potentially all around the world. The deep learning applications could be in all fields that rely on artificial neural networks for tasks such as computer vision, language processing, time series analysis and prediction.”