CNC end milling is a widely used cutting operation to produce surfaces with various profiles. The manufactured parts' quality not only depends on their geometries but also on their surface texture, such as roughness. To meet the roughness specification, the selection of values for cutting conditions, such as feed rate, spindle speed, and depth of cut, is traditionally conducted by trial and error, experience, and machining handbooks. Such empirical processing is time consuming and laborious. Therefore, a combined approach for determining optimal cutting conditions for the desired surface roughness in end milling is clearly needed. The proposed methodology consists of two parts: roughness modeling and optimal cutting parameters selection. First, a machine learning technique called support vector machines (SVMs) is proposed for the first time to capture characteristics of roughness and its factors. This is possible due to the superior properties of well generalization and global optimum of SVMs. Next, they are incorporated in an optimization problem so that a relatively new, effective, and efficient optimization algorithm, particle swarm optimization (PSO), can be applied to find optimum process parameters. The cooperation between both techniques can achieve the desired surface roughness and also maximize productivity simultaneously.