Erin Taylor, Ian Frankenburg, and Eric Peshkin worked with client Dr. John Harer of Geometric Data Analytics to detect abnormal behavior in communications networks. They built models representing networks as nodes for IP addresses and edges for communications between them. Topological and statistical tools were used to study changes in network connectivity and fit probability distributions to the data to identify anomalous events. When comparing test networks to training networks, some nodes in the test networks showed unusual communication patterns. The team identified many potentially anomalous communications and believe further analysis of more comprehensive data sets could yield interesting findings, especially at the network level.