Neural network for online handwriting recognition and its learning method

摘要

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a neural network for online handwriting recognition and a learning method thereof. In the related art, in the field of adaptive learning, such as online handwriting recognition, an algorithm suitable for the characteristics of an application is required. It is not possible to quickly adapt to one user and implement in real time and to implement in real time. Even if it is touched, the volume becomes too large, making it difficult to process in real time. There is a problem that the processing becomes difficult. Therefore, the present invention maintains generality by introducing a fixed representative vector, and enables rapid adaptive learning by generating a representative vector and introducing an effective radius into each vector during adaptive learning, and obtaining a sufficient learning effect with a small amount of data. The recognition time is reduced by using the distance L ₁ , which is the sum of the absolute values of each component difference.