This paper introduces techniques to train feedforward nets to automate ranking and classification tasks. The techniques are denoted mimic nets since the nets can always mimic self-consistent training data. The mimic nets are constructed not for any neurological analogy, but for computational ease and purposeful utility. Mimic nets are designed for problems requiring sensible extrapolation from noiseless training data, and errorless recall of the training data. Linear programming algorithms are utilized to train the net to exactly mimic the expert in all training situations, to identify efficacious features, and to assess the training data. The number of nodes and the number of connections, the structure of the mimic net, are adapted together with weights in the net. The existence of a mimic net for every consistent set of training data is demonstrated.