SVM Support Vector Machine (SVM) for classification task with Stochastic Gradient Descent (SGD) optimization. Architecture allows to use different loss functions (for example, Hinge Loss), regularizations (L1, L2) and models (for example, linear SVM).