Cascading Machine Learning to Attack Bitcoin Anonymity

Abstract

Bitcoin is a decentralized, pseudonymous cryptocurrency that is one of the most used digital assets to date. Its unregulated nature and inherent anonymity of users have led to a dramatic increase in its use for illicit activities. This calls for the development of novel methods capable of characterizing different entities in the Bitcoin network.In this paper, a method to attack Bitcoin anonymity is presented, leveraging a novel cascading machine learning approach that requires only a few features directly extracted from Bitcoin blockchain data. Cascading, used to enrich entities information with data from previous classifications, led to considerably improved multi-class classification performance with excellent values of Precision close to 1.0 for each considered class. Final models were implemented and compared using different machine learning models and showed significantly higher accuracy compared to their baseline implementation.Our approach can contribute to the development of effective tools for Bitcoin entity characterization, which may assist in uncovering illegal activities.

Bitcoin and Cybersecurity: Temporal Dissection of Blockchain Data to Unveil Changes in Entity Behavioral Patterns

“…In [10,12], methods for attacking Bitcoin user anonymity are presented. Both methods use the whole blockchain to create supervised machine learning models and classify Bitcoin entities.…”

“…Both methods use the whole blockchain to create supervised machine learning models and classify Bitcoin entities. In particular, a cascading machine learning model is introduced in [10], which is essentially ensemble learning based on the stacking concept presented in [31]. The idea of the cascading model is to implement a cascade of (weak) classifiers, such that prior classification results can be joined and can be used to enrich a final (strong) classification.…”

“…In this paper, the idea is to join the machine learning model introduced in [10] with the concept of batch analysis applied, for example, in anomalies detection [34,38]. We aim to add a "temporal view" to Bitcoin entity analysis in order to unveil changes in behavioral patterns over time.…”

“…Prior studies have tried to classify entities according to classes representing specific entity behavior within the network [10][11][12]. These techniques usually consider the whole blockchain and thus classification is performed considering all network (macro) dynamics among users.…”

“…Specifically, the present analysis is based on extending cascading machine learning concepts presented in Zola et al [10], which essentially apply ensemble learning and stacking of several classifiers trained with different datasets derived from blockchain data. The idea behind this technique is to classify Bitcoin entities using prior classification information, combine them with new blockchain-derived data, and then compute a final classification to improve predictive performance.…”