Algorithm 2 Non-negative collaborative representation based classifier.
	1:	Normalize $\mathit{X}$ to have $l_2$ norm
	2:	Calculate the collaborative representation coefficients vector $\widehat{\alpha }$ using formulation (1)
$\widehat{\mathit{\alpha }}=argmin\alpha {∥y-X\alpha ∥}_2^2+\lambda {∥\alpha ∥}_2^2$
	3:	Use ReLU function described in Equation (5) to map each collaborative representations to non-negative representation
$f\left(\mathit{x}\right)=\left\{\begin{array}{cc}\mathit{x}& ,\mathit{x}⩾0\\ 0& ,\mathit{x}<0\end{array}\right.$
$\tilde{\mathit{\alpha }}=f\left(\widehat{\mathit{\alpha }}\right)$
	4:	Represent the test sample using non-negative representations described as Equation (6)
$\mathit{y}=\widetilde{{\mathit{\alpha }}_{\mathbf{1}}}{\mathit{x}}_{\mathit{1}}+\widetilde{{\mathit{\alpha }}_{\mathbf{2}}}{\mathit{x}}_{\mathbf{2}}+\widetilde{{\mathit{\alpha }}_{\mathbf{3}}}{\mathit{x}}_{\mathbf{3}}+\dots +\widetilde{{\mathit{\alpha }}_{\mathit{n}}}{\mathit{x}}_{\mathit{n}}$
	5:	Calculate the the residuals between test sample $\mathit{y}$ and representation of each class using Equation (7)
$identity\left(\mathit{y}\right)={argmin}_i{\parallel \mathit{y}-{\mathit{X}}_{\mathit{i}}\widetilde{{\mathit{\alpha }}_{\mathit{i}}}\parallel }_2$