1 December 2021 in Expert Systems with Applications
SASameer Hameed Abdulshaheed
Sameer Hameed Abdulshaheed
12 June 2019 in IEEE Transactions on Vehicular Technology
In this paper we propose Weighted SBL (WSBL) for sparse signal recovery, inspired by the Sparse Bayesian Learning (SBL) method. Unlike SBL, where all hyperparameter priors follow Gamma distributions with identical parameters, in WSBL, the hyperparameters are Gamma distributed with distinct parameters. These parameters, guided by some known weights, give more importance to some hyperparameters over others, thus introducing more degrees of freedom to the problem and leading to better recovery performance. The weights can be determined based on a low-resolution estimate of the sparse vector, for example an estimate obtained via a method that does not encourage sparsity. The choice of the MUltiple SIgnal Classification (MUSIC) estimate as weight is analyzed. Unlike SBL, where the hypeparameters are not bounded, in WSBL there is an upper bound; this make it easy to select a threshold that distinguishes between zero and non-zero elements in the recovered sparse vector, which makes the iterative recovery process converge faster. Theoretical analysis based on variational approximation theory and also simulation results demonstrate that the proposed WSBL results in substantial improvement in terms of probability of detection and probability of false alarm, as compared to existing approaches, such as SBL, Sparse Bayesian Support knowledge (BSN), and Multiple response Sparse Bayesian Learning (MSBL), especially in the low signal to noise ratio regime. The performance of WSBL is evaluated for Direction of Arrival (DOA) in colocated Multiple Input Multiple Output (MIMO) radar.
14 September 2017 in IEEE Transactions on Signal Processing
In sparse signal recovery problems, 11-norm minimization is typically used as an alternative to more complex 10-norm minimization. The range space property (RSP) provides the conditions under which the least 11 -norm solution is equal to at most one of the least 10-norm solutions. These conditions depend on the sensing matrix and the support of the underlying sparse solution. In this paper, we address the problem of recovering sparse signals by weighting the corresponding sensing matrix with a diagonal matrix. We show that by appropriately choosing the weights, we can formulate an 11-norm minimization problem that satisfies the RSP, even if the original problem does not. By solving the weighted problem we can obtain the support of the original problem. We provide the conditions which the weights must satisfy, for both noise free and noisy cases. Although the precise conditions involve information about the support of the sparse vector, the class of good weights is very wide, and in most cases encompasses an estimate of the underlying vector obtained via a conventional method, i.e., a method that does not encourage sparsity. The proposed approach is a good candidate for Electroencephalography (EEG) sparse source localization, where the corresponding sensing matrix has high coherence. The performance of the proposed approach is evaluated via simulations and also via experiments on localizing active sources in the brain corresponding to an auditory task from EEG recordings of a human subject.