Algorithm 1 Multiclass SVM implementation

1:$C_i=y_i{\alpha }_i$ 2:$v\left[N_{Cl}\right]=\{0,0,...,0\}$ 3:function Class = svmpredict($\mathbf{x}$) 4:for i = 1 to $N_S$ do 5:    $K\langle \mathbf{x},{\mathbf{s}}_i\rangle =exp\left(-\frac{\left|\right|\mathbf{x}-{\mathbf{s}}_i{\left|\right|}_2}{2{\sigma }^2}\right)$ 6:end for 7:for j = 1 to $N_{Cl}$ do 8:    for k = j+1 to $N_{Cl}$ do 9:        $f\left(\mathbf{x}\right)={\sum }_{i=1}^{N_S}{C}_{i\left(jk\right)}K\langle \mathbf{x},{\mathbf{s}}_i\rangle -\rho$ 10:        if $f\left(\mathbf{x}\right)>0$ then 11:           $v\left(j\right)$++ 12:        else 13:           $v\left(k\right)$++ 14:    end for 15:end for 16:Class ← index of $max\left(v\right)$