Module 1: Spectral Analysis of Dynamic Graphs

Faculty Advisor: Ambuj Singh

Concepts: Graph Laplacian, Multivariate time series, Classification, Mixture models, Outlier detection.

Research Areas:

Abstract: Spectral analysis of graph structures has been done for a while. What is relatively novel is the use of spectral analysis for understanding graphs with dynamic attributes/values on nodes or edges. This module will build models of dynamic graph behavior, for a single dynamic graph, for a group of similar dynamic graphs, and for groups of dynamic graphs. Students will examine different methods for carrying out spectral analysis, find outliers, and experiment with dynamic graphs from traffic networks, social networks, and information networks.

Active Quarters: 

  • Fall 2015, Steven Munn
  • Spring 2015, Haraldur Tómas Hallgrímsson and Steven Munn