Application and evaluation of Genetic Programming for aimpoint selection

We have applied Genetic Programming techniques to the determination of ship aimpoints from an input vector. The problem we have studied reduces to mapping a multidimensional input vector to a single real output describing one coordinate of the aimpoint. The multidimensional input vector is obtained by segmenting an image of the ship and performing spatial averages. We have found that Genetic Programming techniques can be used to provide a solution to the problem which maps the inputs to the outputs within the quantization level of the problem. For purposes of comparison we have developed artificial neural network and classifier solutions to the same problem based upon C4.5.