Algorithm 2
1.	Input the whitened sample matrix Z and set the maximum number of iteration for every orthogonal component as T.
2.	For t = 1:T
	For i = 1:q
	Judge the condition, as the orthogonal matrix B needs to meet the condition formula BTB = BBT = I, where I is the unit matrix, so each column vector bi in B is required to follow ${\mathit{\text{b}}}_i^T{\mathit{\text{b}}}_i=1,i=1\dots q$ and ${\mathit{\text{b}}}_i^T{\mathit{\text{b}}}_i=0,i,j=1\dots q,i\ne j$, i.e., the norm ║bi (t)║2 = 1 for each iteration t.
	Take the following iteration formula ${\mathit{\text{b}}}_i(t+1)=\text{E}\{\mathbf{\text{z}}f({\mathit{\text{b}}}_i^{\text{T}}(t)\mathbf{\text{z}})\}-\text{E}\{f\prime ({\mathit{\text{b}}}_i^{\text{T}}(t))\}{\mathit{\text{b}}}_i(t)$, where E refers to the mathematical expectation and the activation function is $f(x)=\frac{e^{2x}-1}{e^{2x}+1}$.
	Get the orthogonalization for the ingredients by ${\mathit{\text{b}}}_i(t+1)-\sum _{j=1}^{j-1}<{\mathit{\text{b}}}_i(t+1),{\mathit{\text{b}}}_j>{\mathit{\text{b}}}_j$
	Normalize bi by $\frac{{\mathit{\text{b}}}_i(t+1)}{{\Vert {\mathit{\text{b}}}_i(t+1)\Vert }_2}$
	Continue the iteration until the convergence is achieved.
	End
	End