Multi-layer neural network classification of on-line signatures
N. Mohankrishnan, Wan-Suck Lee, Mark Paulik · 2002
The incorporation of neural network classification strategies to enhance the performance of an autoregressive model-based signature classification system is examined. A multilayer perceptron trained using the back-propagation algorithm is used for classification, and the results obtained from using an extensive database of signatures are presented and compared with those stemming from the use of a conventional maximum likelihood classifier. While there is a definite improvement in the error rates in the signature verification task, accuracies obtained in identification are only marginally better. On the average, the false acceptance and false rejection rates are about 1.7% each, while the identification accuracy is about 97%.