Sparse ECG Denoising with Generalized Minimax Concave Penalty

. 2019 Apr 10;19(7):1718. doi: 10.3390/s19071718

Algorithm 2 FBS algorithm for GMC penalized least squares problem

Input:

x^{(0)}, v^{(0)}, A, A^{T}, λ

Output:

x^{(i + 1)}

1:
Set $ρ = m a x {1, γ / (1 - γ)} \cdot m a x e i g (A^{T} A)$
2:
Set $μ : 0 < μ < 2 / ρ$
3:
Number of iteration: $N_{i t e r}$
4:
for $i = 0 t o N_{i t e r}$ do
5:
$w^{(i)} = x^{(i)} - μ A^{T} (A x^{(i)} - d_{w}) + μ γ A^{T} A (v^{(i)} - x^{(i)})$
6:
$u^{(i)} = v (i) - μ γ A^{T} A (v^{(i)} - x^{(i)})$
7:
$x^{(i + 1)} = s o f t (w^{(i)}, μ λ)$
8:
$v^{(i + 1)} = s o f t (u^{(i)}, μ λ)$
9:
end for
10:
return $x^{(i + 1)}$