Hierarchical Attention Network of Multi-Modal Biometric for a Secure Cloud-Based User Verification

The rise of the Internet has created a need to secure verification for accessing the certain information. The cloud based biometric system can provide a convenient way to realize it. In recent years, various Artificial Intelligence models have been researched for single biometric but rare for multi-modal biometric. This article presents new a multi-model approach as Hierarchical Attention Network (HAN) which is based on four user features i.e palm image, facial image, audio signature, and digital signature. Through the layer-by-layer filtering and improved attention mechanism, HAN can provide a higher verification accuracy, which can provide a more robust and secure user verification service. At the same time, this work also implemented the proposed model on a Cloud-based service with android application. After the evaluation, the HAN method we proposed has been proved to be better than the previous models.