Monday, October 20, 2014

Thesis: Approximate Message Passing Algorithms for Generalized Bilinear Inference, Jason Parker

Here is a new thesis: Approximate Message Passing Algorithms for Generalized Bilinear Inference by Jason Parker

Recent developments in compressive sensing (CS) combined with increasing demands for effective high-dimensional inference techniques across a variety of disciplines have motivated extensive research into algorithms exploiting various notions of parsimony, including sparsity and low-rank constraints. In this dissertation, we extend the generalized approximate message passing (GAMP) approach, originally proposed for high-dimensional generalized-linear regression in the context of CS, to handle several classes of bilinear inference problems. First, we consider a general form of noisy CS where there is uncertainty in the measurement matrix as well as in the measurements. Matrix uncertainty is motivated by practical cases in which there are imperfections or unknown calibration parameters in the signal acquisition hardware. While previous work has focused on analyzing and extending classical CS algorithms like the LASSO and Dantzig selector for this problem setting, we propose a new algorithm called Matrix Uncertain GAMP (MU-GAMP) whose goal is minimization of mean-squared error of the signal estimates in the presence of these uncertainties, without attempting to estimate the uncertain measurement matrix itself. Next, we extend GAMP to the generalized-bilinear case, in which the measurement matrix is estimated jointly with the signals of interest, enabling its application to matrix completion, robust PCA, dictionary learning, and related matrix-factorization problems. We derive this Bilinear GAMP (BiG-AMP) algorithm as an approximation of the sum-product belief propagation algorithm in the high-dimensional limit, where central limit theorem arguments and Taylor-series approximations apply, and under the assumption of statistically independent matrix entries with known priors. In addition, we propose an adaptive damping mechanism that aids convergence under finite problem sizes, an expectation-maximization (EM)-based method to automatically tune the parameters of the assumed priors, and two rank-selection strategies. We then discuss the specializations of EM-BiG-AMP to the problems of matrix completion, robust PCA, and dictionary learning, and present the results of an extensive empirical study comparing EM-BiG-AMP to state-of-the-art algorithms on each problem. Our numerical results, using both synthetic and real-world datasets, demonstrate that EM-BiG-AMP yields excellent reconstruction accuracy (often best in class) while maintaining competitive runtimes and avoiding the need to tune algorithmic parameters. Finally, we propose a parametric extension known as P-BiG-AMP, which recovers BiG-AMP as a special case, that relaxes the assumption of statistically independent matrix entries by introducing parametric models for the two matrix factors. The resulting algorithm is rigorously justified for random affine parameterizations and constructed to allow its use with an even wider class of non-linear parameterizations, enabling numerous potential applications.

Jason is one of the Map Makers.
Join the CompressiveSensing subreddit or the Google+ Community and post there !
Liked this entry ? subscribe to Nuit Blanche's feed, there's more where that came from. You can also subscribe to Nuit Blanche by Email, explore the Big Picture in Compressive Sensing or the Matrix Factorization Jungle and join the conversations on compressive sensing, advanced matrix factorization and calibration issues on Linkedin.

No comments: