To date, there has been considerable growth in the development and application of artificial neural networks. A new approach toward this application, so-called support vector machines (SVM), is proposed. The superior classification and regression abilities of this approach offers an attractive alternative to solve a number of process engineering problems.