The use of probabilistic neural networks to improve solution times for hull-to-emitter correlation problems

Abstract

The probabilistic neural network (PNN) paradigm has been applied successfully to the hull-to-emitter correlation (HULTEC) problem. The PNN is a multilayer feedforward network that uses sums of Gaussian distributions to estimate the probability density function for a training data set. This trained network can then be used to classify new data sets on the basis of the learned probability density functions and, further, to provide a probability factor associated with each class. In the HULTEC applications, the PNN was capable of identifying, with a high degree of accuracy, the emitter of origin of electronic intelligence reports. The data sets were difficult to classify, since regions were separated by nonlinear boundaries and were made up of disjoint multiple and overlapping regions. Tremendous speedup on training was achieved by the PNN implementation compared with the application of backpropagation networks.