1: \begin{abstract}

2: The aim of this paper is to study the convergence of the

3: primal-dual dynamics pertaining to Support Vector Machines

4: (SVM). The optimization routine, used for determining an SVM for classification, is first formulated as a dynamical

5: system. The dynamical system is constructed such that its

6: equilibrium point is the solution to the SVM optimization

7: problem. It is then shown, using passivity theory, that the

8: dynamical system is global asymptotically stable. In other

9: words, the dynamical system converges onto the optimal

10: solution asymptotically, irrespective of the initial

11: condition. Simulations and computations are provided for

12: corroboration.

13: \end{abstract}

14: