Nvidia Deep Learning Institute

Iowa Informatics Initiative (UI3) and ITS-Research Services recently organized the Deep Learning Institute on-campus from July 19-21, 2017. The event included 58 participants spanning four colleges (Medicine, Liberal Arts & Sciences, Engineering, and Public Health). The three day workshop covered Deep Learning Fundamentals (day 1), Deep Learning for Medical Imaging Analysis (day 2), and Deep Learning for Genomic Analysis on the third day (half-day). The sessions consisted of learning activities using Jupyter notebook and the Nvidia’s DIGITS platform with imaging and genomic data on Amazon cloud nodes with Nvidia GPUs. 

The slides for the July 19-21, 2017 Deep Learning Institute at the University of Iowa campus can be found HERE.

Research Services and Nvidia plan to continue to provide opportunities in this area. These include:

  • Research Services will be working with Nvidia to make all material from the course available to University of Iowa students, staff and faculty in the future. (Will require download of the DIGITS platform: https://developer.nvidia.com/digits).
  • Nvidia has offered to provide access to solution architects specializing in specific fields. For example if you have a medical imaging problem that you would like to tackle we can work to arrange a time for you to work one on one with a specialist in this area. Note that this has to very specific aspect that you want to discuss with the Solutions Architect (not train the students for deep learning basics).
  • Nvidia is willing to facilitate meetings with peer institutions that have developed workflows using their tools and technology.
  • Nvidia is willing to collaborate in specific areas where their expertise may be needed. As an example one of the Biomedical Engineering graduate student is working lung image analysis using Nvidia’s cluster.
  • Several opportunities exist to write grants to gain greater access to high performance Nvidia platforms such as the DGX-1 machine learning system (https://www.nvidia.com/en-us/data-center/dgx-1/). If you have research problems that can leverage this type of system and/or are interested in participating in a grant for a DGX-1 or similar infrastructure please contact Jeremy Bockholt at jeremy-bockholt@uiowa.edu.

For further information please contact Sai Ramadugu (saikumar-ramadugu@uiowa.edu).