The Center for Institutional Research Computing’s October seminar will be given by Dr. Larry Holder.
The talk will be held in SPARK 227 on Wednesday, October 10th. Coffee and light refreshments will be served.
Mining Patterns and Anomalies from Dynamic Networks
The onslaught of data continues to challenge our ability to efficiently and effectively extract knowledge from the data. In addition to collecting data about entities and their attributes, most domains also collect information about the relationships between these entities, e.g., sensor networks, computer networks, biological networks and social networks. A natural representation for this information is a graph, and graph mining involves the discovery of knowledge in this relational data. Mining static graphs is challenging enough, but a further challenge is the fact that this information is not static. Entities and their relationships change over time, and we want to incorporate these dynamics into the mining process while avoiding the need to retain all the data. This talk will present work on several dynamic graph mining problems, including pattern discovery and anomaly detection.
