A Neural Network Classifier for the I100 OCR Chip

This paper describes a neural network classifier for the 11000 chip, which optically reads the E13B font characters at the bottom of checks. The first layer of the neural network is a hardware linear classifier which recognizes the characters in this font . A second software neural layer is implemented on an inexpensive microprocessor to clean up the results of the first layer. The hardware linear classifier is mathematically specified using constraints and an optimization principle. The weights of the classifier are found using the active set method, similar to Vapnik's separating hyperplane algorithm. In 7.5 minutes ofSPARC 2 time, the method solves for 1523 Lagrange mUltipliers, which is equivalent to training on a data set of approximately 128,000 examples . The resulting network performs quite well: when tested on a test set of 1500 real checks, it has a 99.995% character accuracy rate. 1 A BRIEF OVERVIEW OF THE 11000 CHIP At Synaptics, we have created the 11000, an analog VLSI chip that, when combined with associated software, optically reads the E13B font from the bottom of checks. This E13B font is shown in figure 1. The overall architecture of the 11000 chip is shown in figure 2. The 11000 recognizes checks hand-swiped through a slot. A lens focuses the image of the bottom of the check onto the retina. The retina has circuitry which locates the vertical position of the characters on the check . The retina then sends an image vertically centered around a possible character to the classifier. The classifier in the nooo has a tough job. It must be very accurate and immune to noise and ink scribbles in the input . Therefore , we decided to use an integrated segmentation and recognition approach (Martin & Pittman, 1992)(Platt, et al., 1992). When the classifier produces a strong response, we know that a character is horizontally centered in the retina. A Neural Network Classifier for the 11000 OCR Chip 939 Figure 1: The E13B font, as seen by the 11000 chip