Sparse Signal Recovery Using Log-Sum Regularization and Adaptive Smoothing
Abstract
We study sparse signal recovery from noisy linear observations using nonconvex log-sum regularization. The log-sum penalty reduces the shrinkage bias of ℓ1 regularization and more closely approximates the ℓ0 regularization, but its nonconvexity can make reconstruction algorithms unstable. To mitigate this instability, we use an adaptive smoothing strategy that determines the smoothing parameter so that the scalar proximal operator remains continuous. Using this proximal operator, we formulate the approximate message passing (AMP) algorithm and derive the corresponding state evolution (SE) recursion. The fixed point of the SE recursion predicts the final mean squared error (MSE) and, in the noiseless limit, the exact-recovery phase transition. To further investigate finite-dimensional reconstruction behavior, we implement an alternating direction method of multipliers (ADMM) algorithm. In the noiseless setting, we find that the empirical success boundary of ADMM closely agrees with the SE-predicted phase transition. In the noisy setting, we observe that AMP closely follows the SE prediction, whereas ADMM qualitatively reproduces the SE-predicted dependence of the final MSE on the regularization parameter. A comparison with ℓ1 regularization shows that log-sum regularization is beneficial in low-density or high-measurement-rate regimes, whereas ℓ1 regularization remains preferable at higher densities and lower measurement rates.
Related
- Scaled-Gradient Hard Thresholding Algorithms for Sparse Signal Recovery
- Rank-One Signal Recovery in Sparse Wishart Noise
- Statistical Foundations for a Google Play User-Review Sentiment Index: Signal Fusion, Shrinkage, Distributional Validation, and Dynamic Smoothing
- Sharp spectral norm concentration and regularization of sparse random tensors
- Reliability-Centered Evaluation of Sparse Longitudinal CT Lesion-Size Forecasting with Conformal Interval Calibration and Gompertz-Inspired Regularization
BibTeX
@misc{morita2026sparse,
title = {{Sparse Signal Recovery Using Log-Sum Regularization and Adaptive Smoothing}},
author = {Keisuke Morita and Masayuki Ohzeki},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2605.10626},
eprint = {2605.10626},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2605.10626},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Sparse Signal Recovery Using Log-Sum Regularization and Adaptive Smoothing
AU - Keisuke Morita
AU - Masayuki Ohzeki
PY - 2026
DA - 2026/09/03
PB - arXiv
DO - 10.48550/arxiv.2605.10626
UR - https://arxiv.org/abs/2605.10626
AB - We study sparse signal recovery from noisy linear observations using nonconvex log-sum regularization. The log-sum penalty reduces the shrinkage bias of ℓ_1 regularization and more closely approximates the ℓ_0 regularization, but its nonconvexity can make reconstruction algorithms unstable. To mitigate this instability, we use an adaptive smoothing strategy that determines the smoothing parameter so that the scalar proximal operator remains continuous. Using this proximal operator, we formulate the approximate message passing (AMP) algorithm and derive the corresponding state evolution (SE) recursion. The fixed point of the SE recursion predicts the final mean squared error (MSE) and, in the noiseless limit, the exact-recovery phase transition. To further investigate finite-dimensional reconstruction behavior, we implement an alternating direction method of multipliers (ADMM) algorithm. In the noiseless setting, we find that the empirical success boundary of ADMM closely agrees with the SE-predicted phase transition. In the noisy setting, we observe that AMP closely follows the SE prediction, whereas ADMM qualitatively reproduces the SE-predicted dependence of the final MSE on the regularization parameter. A comparison with ℓ_1 regularization shows that log-sum regularization is beneficial in low-density or high-measurement-rate regimes, whereas ℓ_1 regularization remains preferable at higher densities and lower measurement rates.
ER -