Applied Math Seminar — Alona Kryshchenko (CSUCI)

Title: Data science and applications in dynamic topic modeling

Abstract:
The shockwaves of the big data boom have thrown into sharp relief the critical need for domain-driven, large-scale data analytic techniques across the fields of, among others, finance, political science, economics, psychology, and medicine. It is not simply the size of data sets that contributes to the extreme challenges of data analysis in these fields, but the inherent complexity of this data. Often this data is multi-modal, with modes representing measurements along different dimensions (e.g., spatial, and temporal dimensions of video data, or word and document dimensions of text corpora data). This data is often naturally formatted as a tensor, a higher-order generalization of a matrix. In this talk, we will explore nonnegative tensor decompositions and their applications in dynamic topic modeling.