基于少样本学习的区块链地址身份推断方法研究

doi:10.3778/j.issn.1002-8331.2405-0014

摘要/Abstract

摘要： 由于区块链的匿名性，钓鱼诈骗、黑客攻击等安全威胁层出不穷，造成了巨大经济损失。区块链身份推断可有效分析匿名地址的交易行为特征，推断地址的身份类型，以加强区块链监管治理。现有的身份推断模型大多依赖大量标记样本进行训练，但是现实中往往难以获取足量的标签地址，对于少量标签的身份类型的检测能力不足。因此，首次提出在少样本环境下的进行区块链地址身份推断，提出了基于少样本学习的区块链地址身份推断方案。为了实现对交易图的深入特征提取，设计了改进的三元组损失函数，该函数在嵌入空间中有效聚集同类样本，同时分离异类样本，显著提升了特征嵌入表示在不同身份类型间的区分度。创新性地提出了一种基于距离加权的原型网络，能够根据样本与均值中心的距离动态调整注意力权重，有效减少异常值和噪声值的干扰，增强类原型计算的准确性。使用度量学习模块将每个样本与各类原型进行距离度量来得到最终的预测结果。这些改进不仅提升了模型在少样本条件下的性能，还增强了模型对未知身份类型推断的泛化能力。实验结果表明，该模型在真实以太坊的交易数据集上测试取得了优异性能，证明了方法的有效性。

关键词: 区块链, 以太坊, 少样本学习, 身份推断

Abstract: Blockchain has attracted wide attention from various fields due to its decentralized and tamper-proof features. However, the anonymity of blockchain has also become a hotbed for attackers to engage in illegal activities, causing huge economic losses every year. Blockchain identity inference can effectively analyze the transaction behavior characteristics of anonymous addresses and infer the identity types of addresses, such as hackers, phishing scams, and so on. Most existing identity inference models rely on a large number of labeled samples for training. However, it is often difficult to obtain sufficient labeled addresses in reality, resulting in insufficient detection ability for identity types with few labels. Therefore, this paper first proposes blockchain address identity inference in a few-shot learning environment. A few-shot learning-based blockchain address identity inference framework is presented. Firstly, to achieve deep feature extraction from the transaction graph, this paper designs an improved triplet loss function that effectively clusters similar samples while separating dissimilar ones in the embedding space, significantly enhancing the discrimination of feature embeddings across different identity types. Secondly, this paper innovatively proposes a distance-weighted prototype network, which dynamically adjusts the attention weights based on the distance between samples and their mean centers, effectively reducing the impact of outliers and noise, and improving the accuracy of prototype calculation. Finally, the metric learning module measures the distance between each sample and various prototypes to obtain the final prediction. These improvements not only enhance performance of the model under few-shot conditions but also improve its generalization capability for inferring unknown identity types. Experimental results show that the model achieves excellent performance on a real Ethereum transaction dataset, demonstrating the effectiveness of the proposed method.

Key words: blockchain, Ethereum, few-shot learning, identity inference

陈彦宇, 黎凯, 付章杰. 基于少样本学习的区块链地址身份推断方法研究[J]. 计算机工程与应用, 2025, 61(12): 311-318.

CHEN Yanyu, LI Kai, FU Zhangjie. Research on Blockchain Address Identity Inference Method Based on Few-Shot Learning[J]. Computer Engineering and Applications, 2025, 61(12): 311-318.

参考文献

[1] ZHU G C, HE D B, AN H Y, et al. The governance technology for blockchain systems: a survey[J]. Frontiers of Computer Science, 2023, 18(2): 182813.
[2] CHEN X X, CHENG Q F, YANG W D, et al. An anonymous authentication and secure data transmission scheme for the Internet of Things based on blockchain[J]. Frontiers of Computer Science, 2024, 18(3): 183807.
[3] NAKAMOTO S. Bitcoin: a peer-to-peer electronic cash system[EB/OL]. (2009-01-12) [2024-05-01]. https://bitcoin.org/bitcoin.pdf.
[4] 毕红亮, 陈艳姣, 伊心静, 等. 基于博弈的加密货币交易市场用户决策优化分析[J]. 软件学报, 2023, 34(12): 5477-5500.
BI H L, CHEN Y J, YI X J, et al. Game-based user decision optimization analysis of cryptocurrency trading market[J]. Journal of Software, 2023, 34(12): 5477-5500.
[5] 王茜, 朱俊伟, 张晓东. 链上链下数据协同下的政务材料共享设计实现[J]. 计算机工程与应用, 2023, 59(20): 333-342.
WANG Q, ZHU J W, ZHANG X D. Design and implementation of materials sharing related to government affairs based on on-chain and off-chain data collaboration[J]. Computer Engineering and Applications, 2023, 59(20): 333-342.
[6] 韦可欣, 李雷孝, 高昊昱, 等. 区块链访问控制技术在车联网中的应用研究综述与展望[J]. 计算机工程与应用, 2023, 59(24): 26-45.
WEI K X, LI L X, GAO H Y, et al. Review and prospect of blockchain access control technology in Internet of vehicles[J]. Computer Engineering and Applications, 2023, 59(24): 26-45.
[7] 蔡元海, 宋甫元, 黎凯, 等. 高判别精度的区块链交易合法性检测方法[J]. 计算机工程与应用, 2024, 60(5): 271-280.
CAI Y H, SONG F Y, LI K, et al. Blockchain transaction legitimacy discrimination with high recognition accuracy[J]. Computer Engineering and Applications, 2024, 60(5): 271-280.
[8] TOYODA K, OHTSUKI T, MATHIOPOULOS P T. Multi-class Bitcoin-enabled service identification based on transaction history summarization[C]//Proceedings of the 2018 IEEE International Conference on Internet of Things (iThings) and IEEE Green Computing and Communications (GreenCom) and IEEE Cyber, Physical and Social Computing (CPSCom) and IEEE Smart Data. Piscataway: IEEE, 2018: 1153-1160.
[9] BREIMAN L. Random forests[J]. Machine Learning, 2001, 45(1): 5-32.
[10] LIN Y J, WU P W, HSU C H, et al. An evaluation of Bitcoin address classification based on transaction history summarization[C]//Proceedings of the 2019 IEEE International Conference on Blockchain and Cryptocurrency. Piscataway: IEEE, 2019: 302-310.
[11] BARTOLETTI M, PES B, SERUSI S. Data mining for detecting Bitcoin Ponzi schemes[C]//Proceedings of the 2018 Crypto Valley Conference on Blockchain Technology. Piscataway: IEEE, 2018: 75-84.
[12] HUANG Y H, WANG H Y, WU L, et al. Understanding (mis)behavior on the EOSIO blockchain[J]. Proceedings of the ACM on Measurement and Analysis of Computing Systems, 2020, 4(2): 1-28.
[13] LIU X, TANG Z Y, LI P, et al. A graph learning based approach for identity inference in DApp platform blockchain[J]. IEEE Transactions on Emerging Topics in Computing, 2022, 10(1): 438-449.
[14] KIPF T N, WELLING M. Semi-supervised classification with graph convolutional networks[J]. arXiv:1609.02907, 2016.
[15] LI Y, CAI Y, TIAN H, et al. Identifying illicit addresses in Bitcoin network[C]//Proceedings of the International Conference on Blockchain and Trustworthy Systems. Singapore: Springer Singapore, 2020: 99-111.
[16] WU J J, YUAN Q, LIN D, et al. Who are the phishers? phishing scam detection on ethereum via network embedding[J]. IEEE Transactions on Systems, Man, and Cybernetics: Systems, 2022, 52(2): 1156-1166.
[17] PEROZZI B, AL-RFOU R, SKIENA S. DeepWalk: online learning of social representations[C]//Proceedings of the 20th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York: ACM, 2014: 701-710.
[18] THEKUMPARAMPIL K K, WANG C, OH S, et al. Attention-based graph neural network for semi-supervised learning[J]. arXiv:1803.03735, 2018.
[19] HAMILTON W, YING Z T, LESKOVEC J. Inductive representation learning on large graphs[C]//Advances in Neural Information Processing Systems (NIPS), 2017: 1-11.
[20] LI S J, GOU G P, LIU C, et al. TTAGN: temporal transaction aggregation graph network for ethereum phishing scams detection[C]//Proceedings of the ACM Web Conference 2022. New York: ACM, 2022: 661-669.
[21] HOCHREITER S, SCHMIDHUBER J. Long short-term memory[J]. Neural Computation, 1997, 9(8): 1735-1780.
[22] 李大秋, 付章杰, 程旭, 等. 基于少样本学习的通用隐写分析方法[J]. 软件学报, 2022, 33(10): 3874-3890.
LI D Q, FU Z J, CHENG X, et al. Universal steganalysis based on few-shot learning[J]. Journal of Software, 2022, 33(10): 3874-3890.
[23] NAKAMURA A, HARADA T. Revisiting fine-tuning for few-shot learning[J]. arXiv:1910.00216, 2019.
[24] FINN C, ABBEEL P, LEVINE S. Model-agnostic meta-learning for fast adaptation of deep networks[J]. arXiv:1703.03400, 2017.
[25] BASU S, HU S, MASSICETI D, et al. Strong baselines for parameter-efficient few-shot fine-tuning[C]//Proceedings of the AAAI Conference on Artificial Intelligence, 2024: 11024-11031.
[26] ALSALEH A M, ALBALAWI E, ALGOSAIBI A, et al. Few-shot learning for medical image segmentation using 3D U-Net and model-agnostic meta-learning (MAML)[J]. Diagnostics, 2024, 14(12): 1213.
[27] VINYALS O, BLUNDELL C, LILLICRAP T, et al. Matching networks for one shot learning[C]//Advances in Neural Information Processing Systems, 2016: 1-9.
[28] SNELL J, SWERSKY K, ZEMEL R S. Prototypical networks for few-shot learning[C]//Advances in Neural Information Processing Systems (NIPS), 2017: 1-11.
[29] SUNG F, YANG Y X, ZHANG L, et al. Learning to compare: relation network for few-shot learning[C]//Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2018: 1199-1208.
[30] KONG L C, DING X L, CHAI X Q, et al. Prototypical graph neural network for few-shot learning[C]//Proceedings of 2021 Chinese Intelligent Systems Conference. Singapore: Springer Singapore, 2021: 580-586.
[31] SOCHER R, CHEN D, MANNING C D, et al. Reasoning with neural tensor networks for knowledge base completion[C]//Advances in Neural Information Processing Systems, 2013.
[32] LIN D, WU J J, HUANG T, et al. Who is who on ethereum? account labeling using heterophilic graph convolutional network[J]. IEEE Transactions on Systems, Man, and Cybernetics: Systems, 2024, 54(3): 1541-1553.
[33] ZHOU J J, HU C K, CHI J L, et al. Behavior-aware account de-anonymization on ethereum interaction graph[J]. IEEE Transactions on Information Forensics and Security, 2022, 17: 3433-3448.
[34] HU S H, ZHANG Z, LUO B Q, et al. BERT4ETH: a pre-trained transformer for ethereum fraud detection[C]//Proceedings of the ACM Web Conference 2023. New York: ACM, 2023: 2189-2197.
[35] LOSHCHILOV I, HUTTER F. Decoupled weight decay regularization[J]. arXiv:1711.05101, 2017.