Algorithm 1: SMOTE

Input: M (number of samples in the minority class), N (% ratio of synthetic minority samples for class balancing), K (number of nearest neighbors), $s_{syn}$ synthetic instance; Choose randomly a subset $\mathcal{S}$ of the minority class data of size $S=\frac{N}{100}M$ (synthetic samples in the minority class) such that the class labels are uniformly distributed; for all$s_i\in \mathcal{S}$do  (1) Find the K nearest neighbors;  (2) Randomly select one of KNNs, called ${\widehat{s}}_i$;  (3) Calculate the distance $d_{i,k}={\widehat{s}}_i-s_i$ between the randomly selected NN ${\widehat{s}}_i$ and the instance $s_i$;  (4) The new synthetic instance is generated as $s_{syn}=s_i+\delta d_{i,k}$ (where $\delta =rand(0,1)$ is a random number between 0 and 1); end for Repeat steps number 2–4 until the desired proportion of minority class is met.