A Survey on Blockchain Anomaly Detection Using Data Mining Techniques

Abstract

With the more and more extensive application of blockchain, blockchain security has been widely concerned by the society and deeply studied by scholars, of which anomaly detection is an important problem. Data mining techniques, including conventional machine learning, deep learning and graph learning, have been concentrated for anomaly detection in the last few years. This paper presents a systematic survey of the blockchain anomaly detection results using data mining techniques. The anomaly detection methods are classified into 2 main categories, namely universal detection methods and specific detection methods, which contain 8 subclasses. For each subclass, the corresponding research are listed and compared, presenting a systematic and categorized overview of the current perspectives for blockchain anomaly detection. In addition, this paper contributes in discussing the advantages and disadvantages for the data mining techniques employed, and suggesting future directions for anomaly detection methods. This survey helps researchers to have a general comprehension of the anomaly detection field and its application in blockchain data.