EnCTN: an enhanced AI-enabled deep learning framework for security enhancement in blockchain transactions

Area of Science:

• Artificial Intelligence
• Blockchain Technology
• Data Security

Background:

• Deep learning offers advanced solutions for Artificial Intelligence (AI)-based Blockchain frameworks.
• Ensuring data reliability, confidentiality, and anonymity in blockchain transactions is crucial.
• Existing methods require enhancement for robust data durability and propagation.

Purpose of the Study:

• To propose a hybrid Blockchain and Deep Learning model for enhanced data durability and transaction analysis.
• To develop a secure deep learning-enabled blockchain transaction model addressing confidentiality and anonymity.
• To improve temporal anomaly detection in blockchain systems.

Main Methods:

• Utilized an enhanced convolutional temporal network (EnCTN) for transaction analysis.
• Employed a sliding window extraction technique for temporal series data.
• Incorporated dilated convolution to capture long-range dependencies.
• Implemented the framework in Ethereum using Python.

Main Results:

• The proposed technique demonstrated improved performance over existing methods in several parameters.
• Achieved enhanced anomaly classification accuracy on the NSL-KDD dataset.
• The framework efficiently detects temporal anomalies with improved computational efficiency.

Conclusions:

• The hybrid Blockchain and Deep Learning approach provides an efficient solution for real-world anomaly detection.
• The EnCTN model significantly enhances data durability and propagation in blockchain systems.
• The framework offers accurate discovery of temporal anomalies and improved computational efficiency.