Title: Geometric Methods for Network Science and Computational Computer Vision
Presenter: Romeil Sandhu

Abstract: In this talk, we will present recent advances in discrete geometry and control theory as applied to network science, computational computer vision, and machine learning. To motivate necessary mathematical ingredients, we begin by revisiting classical vision problems in segmentation, shape analysis, shape registration, and pose estimation. Concepts such as curvature and its connection to not only system robustness, but also in shape reconstruction and related vision tasks will help lay the foundation for a variety of applications in control-based tracking, mesh analysis, and machine learning. From this, we then shift our attention towards how such concepts can be applied in networks to elucidate functional properties of complex systems. Applications in cancer targeted therapy, drug design, congestion, to even the development of economic indicators for financial risk will be discussed.

This talk is designed for a graduate level audience who are seeking PhD research opportunities in computer vision, machine learning, and network science.

Bio: Romeil Sandhu is currently an Assistant Professor at Stony Brook with appointments in Bioinformatics, Computer Science, and Applied Mathematics & Statistics Departments and is the recipient of the 2018 AFOSR YIP Award for work on interactive feedback control for autonomous systems. He first received his B.S. and M.S. degrees from the Georgia Institute of Technology in Electrical Engineering in 2006 and 2009, respectively. Then, he completed his Ph.D. with a focus on control-based computational vision in 2011. Prior to his academic position at Stony Brook, he formed a startup providing government services leading to a successful exit. His current research interest spans a variety of interdisciplinary research areas that includes complex networks, computer vision, mathematical oncology, shape analysis, systems biology, risk analysis, cybersecurity and machine learning with a common underlying interest lying on the intersection of geometry, statistics, and control.