FE-GNN: Feature Enhanced Graph Neural Networks for Account Classification in Ethereum
Keywords: Blockchain, Identity identification, GNN
TL;DR: FE-GNN is a graph neural network based on feature enhancement. It uses the collected transaction data to build a transaction graph and uses graph convolutional networks and graph attention networks to infer the identity of blockchain addresses.
Abstract: Since the birth of the blockchain cryptocurrency trading platform represented by Bitcoin, cryptocurrencies based on blockchain technology have gained widespread attention and accumulated a large amount of transaction data. The analysis of cryptocurrency transactions has become an important research direction with social and economic value, and an important area of blockchain scientific research. Identifying the identity of different cryptocurrency addresses and understanding their behavior is the core challenge to achieve cryptocurrency transaction analysis, otherwise it is difficult to understand blockchain datasets and analyze them with meaningful results. To this end, this paper proposes a blockchain address identity identification method called \textbf{F}eature \textbf{E}nhanced \textbf{G}raph \textbf{N}eural \textbf{N}etworks (FE-GNN). Specifically, a transaction graph is constructed based on the collected transaction data, and graph learning techniques based on graph convolutional networks and graph attention networks are used to infer the blockchain address identity. Experimental results show that the FE-GNN algorithm outperforms previous algorithms.
Primary Area: unsupervised, self-supervised, semi-supervised, and supervised representation learning
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 10507
Loading