The theme for this year's event was Deep Learning for Science. The programme featured talks from leading scientists from across the country who showed the impact of deep learning on a broad set of applications, including precision agriculture, personalized cancer treatment, materials by design, detecting extreme climate events, controlling fusion reactors, managing networks, and tracking particles in advanced physics experiments. The examples went beyond data analysis problems into design and control of experiments and the derivation and refinement of physical models from data, showing how the scientific process and our understanding are being impacted by these methods. Panel discussions from industry and the DOE Labs explored some of the hardware, software, and methods challenges.
"Over the past several years, there have been rapid innovations in deep learning that promise to transform many scientific disciplines and enable new kinds of scientific discovery", stated Sudip Dosanjh, director of Berkeley Lab's National Energy Research Scientific Computing Center (NERSC). "Even so, challenges remain before we will be able to fully realize deep learning solutions in scientific workflows. Our goal is to facilitate this process by creating and supporting interactive opportunities such as the Monterey Data Conference."
The conference was organized by a CSA team that included NERSC's Steve Farrell, who led the team, Dosanjh, Prabhat, Katie Antypas, and Becci Totzke; the Computational Research Division's Talita Perciano, John Shalf, and Esmond Ng; ESnet's Mariam Kiran; and AHSC's Dee Cadena.