Efficient parallel computation of the stochastic MV-PURE estimator by the hybrid steepest descent method

摘要: In this paper we consider the problem of efficient computation stochastic MV-PURE estimator which is a reduced-rank designed for robust linear estimation in ill-conditioned inverse problems. Our motivation result stems from fact that by estimator, while avoiding regularization parameter selection appearing common technique used problems and machine learning, presents computational challenge due to nonconvexity induced rank constraint. To combat problem, propose recursive scheme general form does not require any matrix inversion utilize inherently parallel hybrid steepest descent method. We verify efficiency proposed numerical simulations.