Optimizing Supply Chain Finance with Deep Belief Networks and Proof of Stake Blockchain

Abstract

This research suggests a groundbreaking methodology for the optimization of Supply Chain Finance (SCF) by integrating Deep Belief Networks (DBN) and Proof of Stake (PoS) blockchain techniques. The research collects business details, financial tactics, and supplier information from the Kaggle dataset. This paper uses Tokenization for the process of preprocessing, and it uses Locally Linear Embedding (LLE) method to reduce the dimensionality and in the protection of limited structures. Then, the optimization is done by DBNs, which will improve the method’s accuracy, whereas the PoS secures the data with encryption, decryption, and hashing. Moreover, the method’s effectiveness is evaluated using performance analyses such as computational time, efficiency ratio, error ratio, data authentication, and data management ratio. The proposed DBN-PoS is then compared with some existing methods like Self-adaptive Tasmanian Devil Optimization (SA-TDO), Federated Learning (FL), and Deep Convolutional Neural Network (Deep CNN) to provide a higher accuracy of about 100% and F1-Score of about 99.90%. Furthermore, blockchain integration improves transparency, protects the details of transactions, and safeguards from fraud actions using the SCF system. This research addresses progressing SCF optimization by integrating AI and Blockchain, providing a climbable, well-organized, and protected resolution for real-world entities.