Subspace Methods for Massive and Messy Data
Abstract
This proposal focused on subspace estimation in various modern big-data contexts, where data are massive, streaming, time-varying, and have missing, corrupted, and ill-conditioned data. In the final stages of the project, we also begun an exploration of nonlinear generalizations of subspaces, to provide models that can capture more interesting signal variation.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 12, 2017
- Accession Number
- AD1051284
Entities
People
- Laura Balzano
Organizations
- University of Michigan