University of Nebraska-Lincoln
Statistical Inference for Large Precision Matrices with Applications to Brain Connectivity
Motivated by the importance of precision matrix in many practical applications, this talk considers the statistical inference for precision matrices with ultra high-dimensional time dependent observations. We propose a novel data-driven procedure to construct confidence regions for the precision matrix with application in hypothesis testing and support recovery. Comparing with the existing methods, the proposed procedure imposes weaker structural assumptions on the observed data, which broadens the application scope of this new method. A computationally efficient algorithm is developed to implement the proposed procedure. Applications to brain connectivity by FMRI data is also discussed.
Refreshments at 3:45 pm in Snedecor 2101.