For Deborah Pereg’s PhD in Electrical Engineering at Technion–Israel Institute of Technology, she explored three-dimensional seismic imaging and sparse inversion methods. Seismic imaging reveals subsurface earth layers and geological structures, and can facilitate exploration of mineral deposits and energy sources, or provide geological information in many areas, such as engineering, geothermal energy surveys, or tsunami risk assessment. Dr. Pereg developed simpler and more efficient ways to improve 3D imaging and resolution.
When there is a great deal of data, the system takes a long time to form the images. Dr. Pereg “trained” the imaging system using only a minimal percentage of the seismic data. The system was then able to apply what it learned to the rest of the data, forming the images in much less time.
In the Applied Mathematics program at Yale University for postdoctoral research, Dr. Pereg further explores fast learning, transfer learning, and few-shot learning from a sparse representations point of view, but this time instead of geological, her data is neurological. Her projects include: analyzing high dimensional data from genomics platforms for biomarker discovery and personalized medicine, determining whether complex traits associated with certain common diseases vary across populations with different genetic backgrounds, and investigating the use of pattern analysis methods in brain imaging and genetics for behavioral research.
Dr. Pereg has worked in industry, where she was responsible for algorithm development and implementation.