Algorithm 1. Bayesian Optimization-Support Vector Machine (BO-SVM)

Input: Dataset D, hyper-parameter space Θ, Target score function T($\theta$), max $n^{°}$ of evaluation nmax. Split randomly the D into N folds; one for train set and the other for test set. Build a model m on the train dataset using SVM approach. Choose a starting configuration θ0$ϵ$ Θ. Assess the original score y0 = T(θ0). Initialize S0 = {θ0, y0} While t < maximum number of iterations do For m = 1, …, mmax do Choose a new hyperparameter arrangement θm $ϵ$ Θ by enhancing function Um Θm = ${arg}_{\mathsf{\theta } ϵ \mathsf{\Theta } }$ max Um ($\theta$, St), Analyze H in θm to get a new numerical score ym = T(θm). Strengthen the data ${\mathrm{S}}_{\mathrm{m}}={\mathrm{S}}_{\mathrm{m}-1} \cup$ {θm, ym}. Update the surrogate model. m = m + 1 End for End while Extract optimized hyperparameters. Build SVM model using these tuned hyperparameters from the test data set. Solve the optimization problem, evaluate the accuracy and save it in array. Output: Mean accuracy of classification.