Experiments With A Self-Supervised Adaptive Classification Strategy In On-Line Recognition Of Isolat

Results on a comparison of recognition techniques for on-line recognition of handwritten Latin alphabets are presented. The emphasis is on an adaptive classii-cation strategy introduced in this paper. The classiication strategy is based on compressing or distilling a large database of handwritten characters to a small set of character prototypes. The distillation is performed as a clustering procedure to which the number of clusters are given for each character class. Each user's personal characters are then added to this set as they become available. The classiication decision uses the 1-Nearest Neighbor (1-NN) rule for the distances between the unknown character and the stored prototypes. The distances are calculated using Dynamic Time Warping (DTW). The beneets of the proposed method include the automatic unsupervised learning from user input simultaneously with the normal mode of operation. The presented experiments show that the recognition system is able to adapt to the user's writing style with only a very few { say some tens of { handwritten characters.