NEURAL NETWORK MODELS AND THEIR APPLICATION TO HANDWRITTEN DIGIT RECOGNITION ' Thaddeus

Several neural network paradigms are discussed and their application to the recognition of handwritten digits is considered. In particular, linear auto-associative systems, threshold logic networks, backward error propagation models, Hopfield networks, and Boltzmann machines are considered. An explanation of each technique is presented and its application to dipt recognition is discussed. The tradeoffs of time and space complexity versus recognition accuracy are considered. The objective is to determine the applicability of these techniques to the real-world need of the United States Postal Service for a highaccuracy handwritten digit recognition algorithm. This is especially important in light of the recent interest in these methods. Recognition experiments are presented that were performed on a database of over 10,OOO handwritten digits that were extracted from live mail in a USPS mail processing facility. The time required by each method and their recognition rates are discussed.