Speaker: William Leeb (University of Minnesota)

Title: Matrix Denoising and PCA with Heteroscedastic Noise

Abstract: William will present recent results on the related problems of denoising, covariance estimation, and principal component analysis for the spiked covariance model with heteroscedastic noise. Specifically, he will present an estimator of the principal components based on whitening the noise, and optimal spectral shrinkers for use with these estimated principal components. He will also show new results on the optimality of whitening for principal subspace estimation. This is joint work with Elad Romanov of the Hebrew University.