Author(s)

Jiani Tao

Corresponding Author

Jiani Tao

Abstract

Signature verification is becoming more and more popular in the recent years with growing realistic applications. It serves as an auxiliary personal identity verification method. However, how to establish an effective signature verification system is still an open problem. In this paper, we study utilizing deep learning methods to build such verification system by considering varying feature extractions and different model architectures, ranging from Recurrent Neural Network, Convolutional Neural Network to Transformer. We numerically demonstrate the effectiveness of the built system on benchmark Signature Verification Contest 2004 to get satisfactory evaluation results.

Keywords

Psychological neural network, Digital signature, Verification