BINARY DECISION CLUSTERING FOR NEURAL-NETWORK-BASED OPTICAL CHARACTER-RECOGNITION

A multiple neural network system for handprinted character recognition is presented. It consists of a set of input networks which discrimina te between all two-class pairs, for example ''1'' from ''7'', and an o utput network which takes the signals from the input networks and yiel ds a digit recognition decision. For a ten-digit classification proble m this requires 45 binary decision machines in the input network. The output stage is typically a single trained network. The neural network paradigms adopted in these input and output networks are the multi-la yer perceptron, the radial-bias function network and the probabilistic neural network. A simple majority vote rule was also tested in place of the output network. The various resulting digit classifiers were tr ained on 7480 isolated images and tested on a disjoint set of size 231 40. The Karhunen-Loeve transforms of the images of each pair of two cl asses formed the training set for each BDM. Several different combinat ions of neural network input and output structures gave similar classi fication performance. The minimum error rate achieved was 2.5% with no rejection obtained by combining a PNN input array with an RBF output stage. This combined network had an error rate of 0.7% with 10% reject ion.