Anuj Srivastava: Advancing Analysis of Complex Structured Data

Progress in imaging and storage technology has made cameras, scanners, and video sensors our primary data sources. The wealth of image and video data has revolutionized fields like biometrics, medical diagnostics, bioinformatics, manufacturing, and autonomous navigation. However, dealing with such vast and complex data presents unique challenges, requiring the removal of irrelevant variables to focus on essential structures. This talk explores the convergence of diverse past ideas into a comprehensive approach to handling structured data. It highlights contributions from Kendall's statistical techniques for shape analysis, the Fisher-Rao metric for "information geometry" and its invariance, Grenander's deformable-template approach for moderating complexity, and Silverman and Ramsay's functional analysis for curve data. Recent research has integrated these insights into an ambitious agenda for representing, modeling, and analyzing various data types, leading to significant successes in real-world examples, including COVID data analysis, neuroimaging, computer vision, industrial engineering, and 3D genome analysis.