CAREER: Inference-Driven Data Processing and Acquisition: Scalability, Robustness and Control

The University Of Central Florida Board Of Trustees, Orlando FL

Investigators

Abstract

Ubiquitous sensing has enabled new services and applications that mark every aspect of our lives. However, the ensuing data deluge is a brick wall in face of our abilities to interpret, process and store data. Also, not all the data collected is informative, nor well-attuned to the tasks we care to accomplish. This project introduces new approaches for data processing that hold promise to bring about stunning speedups in the processing of massive data, and explores principled controlled data acquisition paradigms for numerous inference problems. The research activities are expected to advance the theory and practice of data processing and acquisition in emerging cyber-physical systems for civil infrastructure, healthcare, energy, and transportation. The research involves 1) developing, and analyzing the fundamental limits of, transformative subspace-based approaches to data processing that are simultaneously scalable and robust to outliers using subspace pursuit in structure-preserving data sketches, and data-subspace formulations which are invariant to transformations that preserve the underlying low-dimensional data structures, 2) exploring controlled data acquisition in asymptotic regimes of large number of observations with relaxed notions of optimality to unravel the complete structure of optimal design and recognize unifying principles for the design of efficient control policies for a host of controlled inference problems. The practical implications of the fundamental results are studied in the context of structural health monitoring for damage detection and characterization of engineering structures. The educational activities include establishing a new miniature sensing lab, developing an introductory course on learning from big data and sequential analysis, compiling a corpus of intuitive tutorials laying out the core concepts of research results, and mentoring of senior design projects.