Enhancing Credit Risk Assessment in Digital Finance through a Hybrid Deep Learning Model Integrated with Blockchain on the Edge of Things F

This work proposes a credit risk assessment model using deep learning models such as self-attention generative adversarial networks (SA-GAN) and deep multi-layer perceptron (DMLP). Blockchain is used to improve the security aspects of the model by employing Brakerski-Gentry-Vaikuntanathan (BKV) encryption technique. Further, the proposed system is implemented in Edge-of-things network and communications are enabled via LoRaWAN server.本研究提出了一种信用风险评估模型，采用自注意力生成对抗网络（self-attention generative adversarial networks，SA-GAN）与深度多层感知器（deep multi-layer perceptron，DMLP）等深度学习模型。本研究通过区块链（Blockchain）技术结合布拉克斯基-金特里-瓦昆塔纳坦（Brakerski-Gentry-Vaikuntanathan，BKV）加密技术，以提升模型的安全性能。此外，所提出的系统部署于边缘物联网（Edge-of-things）网络中，并通过LoRaWAN服务器实现通信。