Home Department: Computer Science
Mentor: Daniel Rubin, Biomedical Data Science, Radiology, and Medicine (Biomedical Informatics Research)

Deep learning has brought about major breakthroughs in automated medical diagnoses. However, deep learning typically requires a large amount of patient data, so multiple healthcare institutions would have to pool patient data to build a robust deep learning model. There are also many regulatory hurdles to sharing patient data, so Niranjan will develop and deploy data-distributed deep learning methods in which computations are performed on local patient data, thus avoiding the need for data sharing. Such methods will propel collaborative deep learning efforts across multiple healthcare institutions.