As part of an ongoing collaboration, ENCCS and Nvidia corporation co-organized an “AI for Science” Bootcamp event on March 8-9, 2021. Bootcamps such as this one are part of the standard training offering of Nvidia and are designed to give researchers hands-on experience with deep learning (DL) methods through real-life scientific use cases.
On the first day, over 80 participants were first given a broad overview of DL fundamentals and how graphical processing units (GPUs) are used to train, validate and run neural networks, and then worked their way through illustrated Jupyter notebooks to start building their technical skills and understanding. The second day was devoted to a real-life use case based on the classification of hurricanes through satellite images. Participants worked in teams under the guidance of mentors to train convolutional neural networks and optimize their performance on V100 GPUs. While some ran their calculations on an Nvidia training cluster, others used the Alvis cluster in Gothenburg and in the process gained valuable experience in using this Swedish National Infrastructure for Computing (SNIC) resource. The bootcamp mentors deserve a special mention - three experts from SNIC, two from Research Institutes of Sweden AB (RISE), and a handful from Nvidia provided invaluable mentoring to teams of participants and answered questions in a dedicated Slack channel.
Attracting over 80 participants with diverse backgrounds from both academia, industry, and the public sector, training events, such as this bootcamp, are clearly in high demand and we are devoted to further develop the ENCCS AI training program and offer regular training events in partnership with our collaborators.
For more information on the event visit https://enccs.se/events/2021/03/ai-for-science-bootcamp-nvidiaenccs/.