A Non-Convex Optimization Technique for Sparse Blind Deconvolution -- Initialization Aspects and Error Reduction Properties

TL;DR

This paper introduces ALPA, a non-convex optimization algorithm for sparse blind deconvolution that improves accuracy through better initialization and error bounds, especially in speech signal applications.

Contribution

It develops a novel iterative scheme for non-convex sparse blind deconvolution, analyzing initialization effects and demonstrating superior accuracy over existing methods.

Findings

ALPA outperforms state-of-the-art algorithms in deconvolution accuracy.

The method effectively handles non-convexity and regularization in blind deconvolution.

Accurate deconvolution of speech signals into sparse excitation and smooth response.

Abstract

Sparse blind deconvolution is the problem of estimating the blur kernel and sparse excitation, both of which are unknown. Considering a linear convolution model, as opposed to the standard circular convolution model, we derive a sufficient condition for stable deconvolution. The columns of the linear convolution matrix form a Riesz basis with the tightness of the Riesz bounds determined by the autocorrelation of the blur kernel. Employing a Bayesian framework results in a non-convex, non-smooth cost function consisting of an $\ell_2$ data-fidelity term and a sparsity promoting $\ell_p$-norm ($0 \le p \le 1$) regularizer. Since the $\ell_p$-norm is not differentiable at the origin, we employ an $\epsilon$-regularized $\ell_p$-norm as a surrogate. The data term is also non-convex in both the blur kernel and excitation. An iterative scheme termed alternating minimization (Alt. Min.) $\ell_p-\ell_2$ projections algorithm (ALPA) is developed for optimization of the $\epsilon$-regularized cost function. Further, we demonstrate that, in every iteration, the $\epsilon$-regularized cost function is non-increasing and more importantly, bounds the original $\ell_p$-norm-based cost. Due to non-convexity of the cost, the accuracy of estimation is largely influenced by the initialization. Considering regularized least-squares estimate as the initialization, we analyze how the initialization errors are concentrated, first in Gaussian noise, and then in bounded noise, the latter case resulting in tighter bounds. Comparisons with state-of-the-art blind deconvolution algorithms show that the deconvolution accuracy is higher in case of ALPA. In the context of natural speech signals, ALPA results in accurate deconvolution of a voiced speech segment into a sparse excitation and smooth vocal tract response.