Sample-Starved Large Scale Network Analysis
Abstract
In this research project we developed correlation mining methods to answer the following fundamental question about complex networks: What are the fundamental limits on the amount of information that can be inferred about a network from a small number n of indirect empirical observations? In these terms, the overall objective was to develop algorithms and establish performance limits for mining information from correlation networks. The focus was on the sample starved regime arises when the number of variables (columns of the correlation matrix) is of the same order or larger than the number of observations available to estimate or detect patterns in the matrix.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 05, 2016
- Accession Number
- AD1009254
Entities
People
- Alfred O. Hero
- Bala Rajaratnam
Organizations
- Board of Regents of the University of Michigan