Deanonymizing Cryptocurrency With Graph Learning: The Promises and Challenges

Abstract

The world economy is embracing the next generation currency, i.e., cryptocurrencies, which dates back to 2009 when Satoshi Nakamoto made Bitcoin publicly available. Rooted from the nature of decentralization and anonymity of blockchain, the cryptocurrencies have, unfortunately, been leveraged for illicit activities by the criminals. The good news is that typical cryptocurrencies, such as Bitcoin, have to publicly publish their transactions, known as a graph, to retain their ultimate goal of trustless and decentralized transaction verification, which lends law enforcement a means to deanonymizing cryptocurrencies. At meantime, graph learning is an extremely powerful tool to extract the latent features of each vertex in a graph to fulfill various tasks, such as, classifying graph vertices. In this work, we discuss the promises and challenges of exploiting graph learning to deanonymizing cryptocurrencies, which can aid the cyberfighters to circumvent cryptocurrency-based illicit activities.