Detecting Malicious Blockchain Attacks through Flower using Horizontal Federated Learning: An Investigation of Federated Approaches

This research study focuses on the detection of malicious attacks on WEB 3 decentralized applications in blockchain systems using Federated Learning (FL) with the Flower framework. The scope of the study involves evaluating the effectiveness of FL in enhancing the security and reliability of decentralized networks by detecting various types of malicious attacks. A comparative analysis is conducted to assess the performance of FL against traditional machine learning techniques that rely on centralized data storage and direct user data access. The study highlights the limitations of traditional approaches in the context of blockchain systems, emphasizing the need for privacy-preserving collaborative learning mechanisms like FL. By leveraging the Flower framework, FL enables collaborative learning among edge devices without central data storage, ensuring data privacy and improving security. The research findings demonstrate the superiority of FL in terms of privacy preservation and collaborative learning capabilities, establishing its potential for detecting and preventing malicious attacks in blockchain systems. These findings contribute to the wider adoption of blockchain technology by addressing the security concerns associated with decentralized networks.