Tensor decompositions are ubiquitous for analysis and dimensionality reduction of data. They have found application in areas such as neuroscience, market segmentation, hyperspectral image processing, network science, financial portfolio allocation, deep learning, quantum information theory, reinforcement learning, computer vision, drug design, energy demand forecasting, reduced-order models, etc. We will explain the fundamentals of tensor decomposition, focusing on the Tucker, Canonical Polyadic (CP), and Generalized CP (GCP) decompositions, with an emphasis on their application as a tool for unsupervised learning. We will walk through examples of applying tensor decomposition to several illustrative real-world datasets, giving attendees the opportunity to run existing software and analyze results themselves. This tutorial is appropriate for both novices and experts that are interested in better understanding applications of tensor decomposition. The examples will be available for attendees’ future use in their own classroom projects.
MT8: Tensor Decompositions: A Quick Tour of Illustrative Applications
Organizer: Tamara Kolda
MathSci.ai, U.S.
This talk was given at the 2022 SIAM Conference on Mathematics of Data Science in San Diego, California, U.S. Learn more about SIAM Conferences at www.siam.org/conferences/abou...