Title :
Neural approaches for human signature verification
Author_Institution :
Univ. Estadual de Campinas, Sao Paulo, Brazil
Abstract :
This paper describes three neural network (NN) based approaches for on-line human signature verification: Bayes multilayer perceptrons (BMP), time-delay neural networks (TDNN), input-oriented neural networks (IONN). The backpropagation algorithm was used for the network training. A signature is input as a sequence of instantaneous absolute velocity (|v(t)|) extracted from a pair of spatial coordinate time functions (x(t), y(t)). The BMP provides the lowest misclassification error rate among three types of networks
Keywords :
backpropagation; delays; handwriting recognition; multilayer perceptrons; neural nets; Bayes multilayer perceptrons; backpropagation algorithm; human signature verification; input-oriented neural networks; instantaneous absolute velocity; misclassification error rate; network training; neural approaches; spatial coordinate time functions; time-delay neural networks; Backpropagation algorithms; Error analysis; Forgery; Handwriting recognition; Humans; Multi-layer neural network; Multilayer perceptrons; Neural networks; Pattern recognition; Writing;
Conference_Titel :
Document Analysis and Recognition, 1995., Proceedings of the Third International Conference on
Conference_Location :
Montreal, Que.
Print_ISBN :
0-8186-7128-9
DOI :
10.1109/ICDAR.1995.602087
