Support vector machines (SVMs), that utilize a mixture of the L1-norm and the L2-norm penalties, are capable of performing simultaneous classification and selection of highly correlated features. These SVMs, typically set up as convex programming problems, are re-formulated here as simple convex...