A New Method for Constructing Digital Signature With Generative Adversarial Neural Network
Digital signature is an encryption mechanism used to verify the authenticity and integrity of message, which has higher complexity and security than traditional handwritten signature. However, the two main challenges of digital signature are security and computing speed. It then imposes a problem - how to quickly verify and sign digital signatures under the premise of ensuring security. To tackle the above challenges, we propose a new method for constructing digital signatures with generative adversarial neural network, which allows fuzzy or fault-tolerant signatures to improve the practicability of digital signatures. In addition, our method can reduce time consumption of signing and verifying digital signatures through using adversarial neural network. To strengthen the security of our method, we use adversarial neutral networks in GAN to simulate various type of uncertain and unfixed attacks and to implement adjusting parameters adaptively.
The convolution neural network is implemented in Winows10 system. We use Python3.7, Pytorch framework and Anaconda to build environment.