Time-varying Networks Estimation and Dynamic Model Selection

Time-varying Networks Estimation and Dynamic
Model Selection
Annie Qu, University of Illinois at Urbana-Champaign
In many biomedical and social science studies, it is important to identify and predict the
dynamic changes of associations among network data over time. We propose a varyingcoefficient model to incorporate time-varying network data, and impose a piecewisepenalty function to capture local features of the network associations. The advantages of
the proposed approach are that it is nonparametric and therefore flexible in modeling
dynamic changes of association for network data problems, and capable of identifying the
time regions when dynamic changes of associations occur. To achieve local sparsity of
network estimation, we implement a group penalization strategy involving overlapping
parameters among different groups. However, this imposes great challenges in the
optimization process for handling large-dimensional network data observed at many time
points. We develop a fast algorithm, based on the smoothing proximal gradient method,
which is computationally efficient and accurate. We illustrate the proposed method through
simulation studies and children's attention deficit hyperactivity disorder fMRI data, and
show that the proposed method and algorithm efficiently recover dynamic network
changes over time. This is joint work with Xinxin Shu and Lan Xue.
The Johns Hopkins Bloomberg School of Public Health
Department of Biostatistics, Monday, October 13, 2014, 12:15-1:15
Room W3008, School of Public Health (Refreshments: 12:00-12:15)
Note: Taking photos during the seminar is prohibited
For disability access information or listening devices, please contact the Office of Support
Services at 410-955-1197 or on the Web at www.jhsph.edu/SupportServices. EO/AA