Cluster GAN-based model for Signature Generation and Verification

Signatures serve as a popular method of personal identity verification in various sectors such as finance and government. Signature verification systems utilize biometric samples to differentiate between authentic and falsified signatures. Although deep learning presents a promising avenue for creating such systems, the requirement of large signature datasets presents a challenge. Current datasets are restricted in size due to privacy concerns. To overcome this limitation, a model that employs cluster generative adversarial networks is proposed. This model utilizes unconditional GANs to expand the signature dataset, which in turn can be used for deep learning-based signature verification. This model can also be potentially utilized in other domains that require privacy.