Abstract:Sparse signal recovery is one of the most fundamental problems in various applications, including medical imaging and remote sensing. Many greedy algorithms based on the family of hard thresholding operators have been developed to solve the sparse signal recovery problem. More recently, Natural Thresholding (NT) has been proposed with improved computational efficiency. This paper proposes and discusses convergence guarantees for stochastic natural thresholding algorithms by extending the NT from the deterministic version with linear measurements to the stochastic version with a general objective function. We also conduct various numerical experiments on linear and nonlinear measurements to demonstrate the performance of StoNT.

What problem does this paper attempt to address?

The paper attempts to address the problem of improving the computational efficiency and accuracy of algorithms in sparse signal recovery. Specifically, the paper proposes two new Stochastic Natural Thresholding algorithms (StoNT and Stochastic Natural Thresholding Pursuit, StoNTP), aiming to solve the sparse signal recovery problem by extending the natural thresholding method in deterministic linear measurements to more general stochastic versions. These algorithms not only provide convergence guarantees theoretically but also demonstrate through numerical experiments that StoNTP has faster convergence speed and higher recovery accuracy compared to existing natural thresholding algorithms (such as NTP) when dealing with linear and nonlinear measurements. The main contributions of the paper include: 1. **Proposing new algorithms**: The paper introduces two algorithms, StoNT and StoNTP, which are stochastic versions of the natural thresholding algorithms (NT and NTP), suitable for a broader range of objective functions. 2. **Theoretical analysis**: The paper provides convergence analysis of these new algorithms in solving the sparse signal recovery problem, particularly in the case of linear measurements. 3. **Numerical experiments**: Through extensive numerical experiments, the paper validates the performance of the StoNTP algorithm under different parameter settings, especially demonstrating its superiority compared to the traditional NTP algorithm. In summary, by introducing stochastic natural thresholding algorithms, the paper provides a new effective method for sparse signal recovery, particularly suitable for scenarios involving large data volumes and high-dimensional data.

Stochastic Natural Thresholding Algorithms

Newton-Type Optimal Thresholding Algorithms for Sparse Optimization Problems

Stochastic IHT With Stochastic Polyak Step-Size for Sparse Signal Recovery

Newton-Step-Based Hard Thresholding Algorithms for Sparse Signal Recovery

Stochastic Iterative Hard Thresholding for Low-Tucker-Rank Tensor Recovery

Projective Iterative Hard Thresholding Algorithm for Sparse Signal Recovery

Stochastic Greedy Algorithms For Multiple Measurement Vectors

Randomized Iterative Hard Thresholding for Sparse Approximations

Normalised Iterative Hard Thresholding; guaranteed stability and performance

Stochastic Recursive Gradient Support Pursuit and Its Sparse Representation Applications

Fast thresholding algorithms with feedbacks for sparse signal recovery

Heavy-Ball-Based Hard Thresholding Algorithms for Sparse Signal Recovery

Stochastic Recovery Of Sparse Signals From Random Measurements

Data-Time Tradeoffs for Optimal k-Thresholding Algorithms in Compressed Sensing

Projective Hard Thresholding Pursuit for Nonnegative Sparse Recovery

Non-negative Sparse Recovery via Momentum-Boosted Adaptive Thresholding Algorithm

Distributed Sparse Recursive Least-Squares over Networks

Iterative Difference Hard-Thresholding Algorithm for Sparse Signal Recovery.

Nonlinear Functions in Learned Iterative Shrinkage-Thresholding Algorithm for Sparse Signal Recovery

Double overrelaxation thresholding methods for sparse signal reconstruction

Linear Convergence of Stochastic Iterative Greedy Algorithms With Sparse Constraints