Ethereum Transaction Anomaly Detection by Integrating Machine Learning Models and Fuzzy Networks for Enhanced Security and Real-Time Monitoring

This research aims at developing an R&D for a hardiness relay alert system which includes applying machine learning and fuzzy logic networks for detecting Ethereum transaction match failures in real time and enhanced Ethereum blockchain security. Since, for example, value, gas price, and smart contract interactions, the system for analyzing transactions correlates with concrete intrinsic characteristics, it selects transactions with suspicious or malicious behavior. This research employs logistic regression, support vector machines (SVM) decision trees and random forests optimized by grid search. On the other hand, the issue of uncertainty and false alarms is solved through the use of fuzzy logic, where fuzzy membership functions place attributes of transactions into linguistic hobbled variables such as ‘low’, ‘medium’ and ‘high’. This descriptive research proves that integrating fuzzy logic with machine learning enhances the approach of mediating anomalies compared and emphasizes that it is better than the rules-based approach. The study concludes with a detailed analysis of the models' effectiveness, illustrated through various graphical representations of the decision-making processes and membership functions, highlighting the system's potential for real-time deployment in securing blockchain networks.