Models and Metrics for Composite Socio-Spatial Networks

Abstract

The objective of this proposal is to build a framework to analyze and build metrics for composite networks. Current analysis techniques are based on projections which inevitably lead to information loss and do not have the flexibility to discover hidden patterns in the network. Similarly, traditional network analysis tools (such as centrality measures) cannot be applied on composite networks. The approach we propose to examine with this project for the analysis of composite networks is to develop a framework that utilizes tensor analysis and decomposition to discover higher order dependencies within composite networks. The framework will include the multiple types of nodes and/or edges of the composite network, as well as their evolution over time. Apart from this tensor-based framework, we will also use statistical analysis techniques to extend our preliminary research into network measures developed for composite networks as well as use and extend these techniques for other metrics.

Document Details

Document Type
DoD Grant Award
Publication Date
Jan 12, 2017
Source ID
W911NF1510599

Entities

People

  • Konstantinos Pelechrinis

Organizations

  • Army Contracting Command
  • United States Army
  • University of Pittsburgh

Tags

Fields of Study

  • Computer science

Readers

  • Neural Network Machine Learning.
  • Structural Health Monitoring of Composite Structures.