Sequential change-point detection over dynamic networks
Streaming data over dynamic networks is ubiquitous nowadays, such as those generated by physical sensors and those by people (e.g., over social networks). A fundamental question is how to monitor the system to detect and localize any change. The challenges come from the high-dimensionality of the data, dynamic background, as well as, complex dependence structure between the data streams. One has to come up with novel statistics which capture the nature of the change. In this talk, Professor Xie will present recent work on two related problems: detect changes over discrete events streams, detect changes over a dynamical network by pairwise comparison. The first problem arises from studying social networks where the user activities are modeled as a networked Hawkes process. The second problem stems from studying dynamic power networks. She has established likelihood ratio based statistics and studied their theoretical properties, as well as, demonstrated their promise in working with real data.
Yao Xie joined Georgia Institute of Technology as an Assistant Professor in the H. Milton Stewart School of Industrial & Systems Engineering in 2013. Prior to that, she worked as a Research Scientist at Duke University in the Department of Electrical and Computer Engineering, after receiving her Ph.D. in Electrical Engineering (minor in Mathematics) from Stanford University in 2011. She is a elected member of IEEE Machine Learning for Signal Processing Technical Committee (MLSP TC) for a term 2016-2019. She is interested in signal processing, sequential analysis (e.g. change-point detection), and machine learning, and have been working on applications in sensor networks, social networks, imaging, and wireless communications. More information can be found at http://www2.isye.gatech.edu/~yxie77/