Description
TL;DR: A small machine-learning experiment that treats unusual graph behavior as a proxy for suspicious Bitcoin users or transactions when no reliable labels exist.
About This Document
Thai T. Pham and Steven Lee build two graphs from Bitcoin transaction data: one with users as nodes and another with transactions as nodes. They extract graph features and apply three unsupervised approaches—k-means clustering, Mahalanobis distance, and an unsupervised support-vector machine—to identify outliers. The authors frame anomaly detection as a practical response to unlabeled financial-network data rather than claiming that every outlier is fraudulent.
Why Bitcoin People May Care
Bitcoin's public ledger makes large transaction graphs available, but it does not supply a clean list of criminal and innocent actors. This paper shows both the appeal and the weakness of behavior-based screening: anomaly can prioritize investigation, yet it is only a proxy for wrongdoing. Read the results as a 2016 research experiment, not a current forensic standard, compliance rule, or accusation against any address.
What You Receive
Purchase reveals the verified preserved PDF location for this document.
Document Details
- Title: Anomaly Detection in the Bitcoin Network Using Unsupervised Learning
- Author / organization: Thai T. Pham and Steven Lee
- Year: 2016
- Language: English
- Document type: Machine-learning research paper
- Pages: 5
CacheRat provides researched source-location and document-identification information. CacheRat does not claim ownership of the underlying third-party documentation.
