Modeling High-Dimensional Time Series


Xiaoping Shi, Department of Mathematics and Statistics, Thompson Rivers University


2018.05.28 16:00-17:00


Middle Lecture Room, Math Building


Modeling high-dimensional time series is necessary in many fields such as neuroscience, signal processing, network evolution, text analysis, and image analysis. Such a time series may contain unknown multiple change-points. For example, the time of cell divisions can be accessed using an automatic embryo monitoring system by a time-lapse observation. When a cell divides at some time point, the distribution of pixel values in the corresponding frame will change, and hence the detection of cell divisions can be formulated as a multiple change-point problem. In this talk, a powerful graph-based change-point detection is introduced.