This second part will focus on specific Pattern Recognition and Machine Learning algorithms. First, we will study one the most famous non parametric algorithms: the k-Nearest Neighbor classifier. We will show that this method, based on the saying “Birds of a feather flock together“, is simple, works well and is theoretically well founded. The main drawback of this method comes from its time and space complexity. Some simple strategies will be presented to overcome those issues. Then, a special focus will be made on Support Vector Machines (SVM) which is nowadays one the most popular machine learning methods. SVM relies on the notion of margin maximization and makes use of the so-called kernel trick that allows us to learn linear separators in high dimensional spaces to deal with non linear problems. We will present in this module the notions of Lagrange multipliers, primal and dual forms, the kernelization of the dual problem, and the L1-norm soft margin SVM.