Обнаружение нелегальных транзакций биткойна при помощи графовых нейронных сетей

Abstract

This article discusses possible methods for solving the problem of classifying bitcoin transactions into licit and illicit ones. The proposed options for solving this problem of anti-money laundering analytics based on graph neural networks with different architecture. The results of computational experiments using data from the Elliptic Data Set are presented.