Comparisons of a neural network and a nearest-neighbor classifier via the numeric handprint recognition problem

A comparison is made of two techniques for recognizing numeric handprint characters using a variety of features including 2D fast Fourier transform coefficients, geometrical moments, and topological features. A backpropagation network and a nearest neighbor classifier are evaluated in terms of recognition performance and computational requirements. The results indicate that for complex problems, the neural network performs comparably to the nearest-neighbor classifier while being significantly more cost effective.