AI summary

This paper introduces a neural network architecture using backpropagation for handwritten zip code recognition, achieving state-of-the-art results on a dataset of 9298 segmented numerals. The model integrates constraints through its architecture, demonstrating the ability of backpropagation networks to handle large amounts of low-level information.

Abstract

The ability of learning networks to generalize can be greatly enhanced by providing constraints from the task domain. This paper demonstrates how such constraints can be integrated into a backpropagation network through the architecture of the network. This approach has been successfully applied to the recognition of handwritten zip code digits provided by the U.S. Postal Service. A single network learns the entire recognition operation, going from the normalized image of the character to the final classification.