Cryptocurrency money laundering is the equivalent of a serious offense since it assists criminals to conceal their actions, in addition to disrupting markets and the financial sector. Researchers are developing effective Anti-Money Laundering (AML) systems to combat this vice. These systems are beneficial to the society because they mitigate the damage committed by crime. This paper explores the possibility of detecting Bitcoin transactions using the Graph Neural Networks (GNNs). Particular types of GNNs, including Graph Convolutional Networks (GCNs), Graph Attention Networks (GATs), Chebysev convolutional neural networks, and Graph SAGE networks are used in the study. We tested various sets of features after examining the data. We discovered that GNN convolutions with a final linear layer and skip connections are more effective in case of the best results, and with Chebyshev and GATv2 convolutions.
Pranathi et al. (Wed,) studied this question.