UC Irvine

Deep Learning has fundamentally shifted the focus of Machine Learning research and Computer Vision in the past decade, largely enabled by advances in hardware acceleration, with single consumer grade GPUs in 2021 outperforming supercomputers from 20 years prior. This work is dedicated to explore a multitude of technological and scientific applications of deep learning in the physical sciences and Medicine. Especially in the case of Medicine, a major challenge is data collection, resulting in typically small data sets due to the difficulty of collecting large amounts of patient data and the significant privacy concerns and policies against sharing the data. Nevertheless, we show that, and how, deep learning can be used to create surprisingly good solutions to problems that were not adequately solvable with prior approaches, even in cases where training data is scarce.