A0284
Title: The asymptotic distribution of the canonical variable in high-dimensional canonical correlation analysis
Authors: Xiaozhuo Zhang - Northeast Normal university (China) [presenting]
Abstract: The asymptotic distribution of the canonical variable in the high-dimensional CCA problem is introduced. The canonical variable is actually the eigenvector of the canonical correlation matrix. Hence, the direction of the canonical variable is described by the cosine of the angle between the eigenvectors, which correspond to the spike eigenvalues of the population and sample canonical correlation matrix, respectively, and give the asymptotic distribution of the cosine of the above angle. Simultaneously, in order to relax the assumption of partial normality, the substitution theorem of the quadratic form is also established.
