BP nets applied to ISAR object recognition

The performance of backpropagation (BP) neural classifiers for inverse synthetic aperture radar (ISAR) object recognition problems has been compared to that of a linear classifier and a nearest-neighbor classifier trained with the same data. The experimental results show that the error (misclassification) rate of the linear classifier is about twice that of the BP classifier, and the error rate of the BP classifier is about twice that of the nearest-neighbor classifier.<>