AUTOMATIC FEATURE GENERATION FOR HANDWRITTEN DIGIT RECOGNITION

An automatic feature generation method for handwritten digit recogniti on is described. Two different evaluation measures, orthogonality and information, are used to guide the search for features. The features a re used in a backpropagation trained neural network. Classification ra tes compare favorably with results published in a survey of high-perfo rmance handwritten digit recognition systems. This classifier is combi ned with several other high performance classifiers. Recognition rates of around 98% are obtained using two classifiers on a test set with 1 ,000 digits per class.