Most current technologies in modeling time varyingprocesses aim to adapt a static model over time in whathas become to be known as continuous learning. Wepropose here a different approach to the same problemdomain that of including the time explicitly in themodeling. An example implementation of this strategy isgiven in form of a multilayer perceptron with explicit timeinput. The performance of this approach is evaluating on abenchmark that was constructed to illustrate typicalproblems in industrial applications.