Hybrid Deep Learning Approach Utilizing RNN and GRU for Fraud detection in Bitcoin cryptocurrency on the Blockchain platform

The rise in financial fraud cases, despite technological advancements, poses a significant challenge due to data inaccessibility and privacy concerns. This study proposes a novel approach combining blockchain technology with deep learning and machine learning algorithms for fraud detection. Blockchain's secure nature is integrated into financial systems, yet fraud continues to increase annually. The suggested model utilizes a graph constructed from sequential Bitcoin transactions, extracting features for classification within the blockchain network. Machine learning methods are employed to train on fraudulent transaction patterns and predict new transactions. By combining blockchain technology with machine learning, the model aims to identify fraudulent transactions effectively. Neural network algorithms with short-term memory are utilized for transaction pattern classification and prediction. Feature selection is done using the XGBoost algorithm, and deep recurrent neural networks with gate mechanisms are used for feature extraction. Model accuracy is measured through precision and area under the curve calculations. This innovative approach seeks to enhance fraud detection efficacy in financial systems by leveraging the strengths of blockchain, deep learning, and machine learning technologies.

کلیدواژه ها:

Bitcoin ، Blockchain ، Fraud detection ، Deep learning