Inertial projection neural network for nonconvex sparse signal recovery with prior information
L1 Lq 0 < q < 1 L1−2 Lq L0 q Mathematically, the above problem can be modeled as the recovery of a given vector from its few measurements , where with is denoted by the measurement matrix. A direct way is to solve an problem. However, the problem is in general NP-hard. To overcome this difficulty, many new methods were proposed successively, such as the classical method [1], method with [2], the method [3], etc. Among these methods, the method is very attractive due to its good property in approaching the method when tends to 0.
Paper
Full text
Inertial projection neural network for nonconvex sparse signal recovery with prior information
Semantic Scholar · Computer Science · 2023
Abstract
L1 Lq 0 < q < 1 L1−2 Lq L0 q Mathematically, the above problem can be modeled as the recovery of a given vector from its few measurements , where with is denoted by the measurement matrix. A direct way is to solve an problem. However, the problem is in general NP-hard. To overcome this difficulty, many new methods were proposed successively, such as the classical method [1], method with [2], the method [3], etc. Among these methods, the method is very attractive due to its good property in approaching the method when tends to 0.