Writer adaptation of online handwritten recognition using Adaptive RBF Network

In this paper, we have proposed an Adaptive Radial Basis Function Network (ARBFN) based writer adaptation for online handwriting recognition. We used RBF Network for both training (writer independent recognition system) and adaptation (writer dependent recognition system) using incremental learning. The adaptation is performed by either adding a center to the present network or modifying the weights of the existing prototype vectors using standard LMS gradient descent, whenever a misclassification is reported. ARBFN is evaluated on devanagari characters and numerals, collected in house. We compare our work with four other approaches using a similar concept of incrementally building the RBF Network. However, these four approaches use RBF Network only for writer adaptation. We also propose a change in the adaptation algorithm when a center is being formed and show that it improves the performance by an average of 5% and also forms lesser centers than the previous approaches. Performance evaluation shows our system improves the error rate reduction on an average by 51% and 28% on devanagari characters and numerals data-set respectively, over the existing approaches.