Toward Anomaly Detection in the Blockchain Using Numerical Association Rule Mining

With the global increase in the popularity of cryptocurrencies, the need for anomaly detection and fraudulent behavior is reaching an all-time high. In our paper, we propose a novel method of anomaly detection with the use of Numerical Association Rule Mining with Differential Evolution. The experiment was conducted by using the Dogecoin blockhain, and the dataset contained all of the transactions from one month. Our results contained 303 rules, with the best fitness function value of 0.8.