COMPARISON OF CRISP AND FUZZY CHARACTER NEURAL NETWORKS IN HANDWRITTEN WORD RECOGNITION

Experiments comparing neural networks trained with crisp and fuzzy des ired outputs are described. A handwritten word recognition algorithm u sing the neural networks for character level confidence assignment was tested on images of words taken from the United States Postal Service mailstream. The fuzzy outputs were defined using a fuzzy k-nearest ne ighbor algorithm. The crisp networks slightly outperformed the fuzzy n etworks at the character level but the fuzzy networks outperformed the crisp networks at the word level. This empirical result is interprete d as an example of the principle of least commitment.