Email: ankurt AT uw dot edu
Office: TLB 307C
Phone: 253-692-4806
Box 358426, 1900 Commerce St,
Tacoma, WA 98402
For AI to be truly acceptable in healthcare, it has to be trustworthy and open. i.e. it has to be Assistive not Artificial. Explanability is an important component of this. So is reducing Bias in machine learning. Together with colleagues from industry and academia my work is currently focused on Responsible AI and its applications in Healthcare.
Transformational Health Care is possible if clinical and operational decisions can be supported by a confluence of data, machine learning, and cloud computing. My group at the Center for Data Science works on a variety of initiatives in healthcare analytics.
My early training was in using CV and PR techniques for handwriting recognition and novelty detection in streams. I continue to be active in these areas and contribute to efforts in video stream analytics. Notable efforts included distributed algorithms for scalable object tracking, obstacle avoidance for low-powered drones, and dietary and volumetric visual analysis of food.
I am interested in the scalability of machine learning pipelines. I collaboratively explore selected topics at the intersection of machine learning and databases including distributed query processing and spatial databases.