A Unified Fractional Regularization Framework for Sparse Recovery
The paper introduces a unified fractional regularization framework ($\ell_1/\ell_p^q$ model) for sparse signal recovery, generalizing existing sparsity-promoting methods. It proves that critical points of this new model are equivalent to those of subtractive $\ell_1-\alpha\ell_p$ problems, providing a unified view. A new sufficient recovery condition under RIP is established, and an efficient Majorization–Minimization algorithm with KL convergence is developed.