Dr. Kevin Xu (MS PhD EE:Systems '09 '12) took first prize at the Challenge Problem competition sponsored by the 2013 International Conference on Social Computing, Behavioral-Cultural Modeling, and Prediction Conference (SBP 2013).
The event, now in its second year, was initiated "to encourage researchers to combine the power of big data and the power of systems thinking in the field of social computing, behavioral-cultural modeling, and prediction."
The challenge problem required the participants in the competition to discover ways to interpret given data sets in a way that could be used to predict future social behavior. The ultimate goal is to make mobile phones the essential tool for conducting social science research, and to support mobile commerce with a solid social science foundation.
According to Dr. Xu, data about the daily activities of four different groups of individuals were collected using smartphones as well as other wearable human sensors, and included information about their physical location at about 5 minute intervals. The data was collected by the MIT Human Dynamics Lab.
A previous study showed that a person's location at any given time is extremely predictable. However, it has not been shown to what extent someone's social interactions could be predicted based on data collected about them. "Understanding the predictability of social interactions has profound implications ranging from indirect marketing to prediction of disease propagation," stated Dr. Xu.
His study of the data showed that an individual's social interactions are much less predictable than their physical locations. To deal with that uncertainty, he showed that a simple Markov chain model could serve as a reasonable approximation for social interactions over time. He stated that "the simplicity of such a model enables us to perform theoretical analyses of interaction networks with real-world justifications."
Dr. Xu presented his findings at the SBP conference along with the 2nd and 3rd place winners.
Dr. Xu is currently a a Senior Research Scientist in the Computational Intelligence Laboratory at 3M. He performs research and development in statistical signal processing and machine learning with applications to security and health care, among other fields. His advisor at U-M was Alfred O. Hero, R. Jamison and Betty Williams Professor of Engineering.
April 25, 2013
EECS/ECE Communications Coordinator
firstname.lastname@example.org or 734-936-2965