Transaction Graph Analysis for Bitcoin Address Classification - Traditional Supervised Machine Learning and Deep Learning Methods - Thesis

Aug 4, 2026

Upload by Ezra

Transaction Graph Analysis for Bitcoin Address Classification focuses on Bitcoin address classification and clustering for law enforcement and regulatory compliance use cases. The work proposes a machine-learning framework that assigns a Bitcoin address to predefined entity classes or to a specific company. Five coarse-grained classes are defined and 180 companies are targeted for fine-grained identification. Contributions include a 3M-address labeled dataset with engineered feature vectors, comparative experiments across techniques, and two classifier families (boosted trees and neural deep learning). Results report strong F1 performance and company linking accuracy.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages73

Reading Time3 h 4 min

AccessFree to download

Introduction

Background

Bitcoin transactions graph behavior

Feature explanation

Classifier methods

Literature review

Classification framework

Creating dataset

Analysis existing methods

Our classifier and evaluation

Deep learning model

Implementation

Transaction Graph Analysis for Bitcoin Address Classification: Traditional Supervised Machine Learning and Deep Learning Methods Seyedarash Saeidimanesh A Thesis In the Concordia Institute for Information Systems Engineering Presented in Partial Fulfillment of the Requirements For the Degree of Master of Applied Science (Information Systems Security) at Concordia University Montral, Qubec, Canada March 2024 © Seyedarash Saeidimanesh, 2024 CONCORDIA UNIVERSITY School of Graduate Studies This is to certify that the thesis prepared By: Seyedarash Saeidimanesh Entitled: Transaction Graph Analysis for Bitcoin Address Classification: Tradi tional Supervised Machine Learning and Deep Learning Methods and submitted in partial fulfillment of the requirements for the degree of Master of Applied Science (Information Systems Security) complies with the regulations of this University and meets the accepted standards with respect to originality and quality. Signed by the Final Examining Committee: Chair Dr. A. Youssef Examiner Dr. J. Clark Supervisor Dr. I. Pustogarov Approved by Dr. J. Yan, Graduate Program Director Dr. M. Debbabi, Dean of the Gina Cody School of Engineering and Computer Science Abstract Transaction Graph Analysis for Bitcoin Address Classification: Traditional Supervised Machine Learning and Deep Learning Methods Seyedarash Saeidimanesh In this thesis, we consider the problem of Bitcoin address classification and clustering, common in the domains of law enforcement and regulatory compliance. We build a machine learning-based classification framework which is able to attribute a Bitcoin address to one of the predefined classes or to a specific company. We consider five distinct classes for coarse-grained classification: cryptocurrency exchanges, online marketplaces, mining pools, fundraising/charity platforms, and gambling; and 180 companies for fine-grained classification. Classes and the companies were selected so that they represent a broad spectrum of entities and activities within the Bitcoin ecosystem. This thesis has three main contributions. First, due to the lack of publicly available datasets suitable for testing machine-learning classification algorithms, we create our own labeled dataset consisting of 3M Bitcoin addresses (from 2016-2022), with each Bitcoin address assigned a readyto-use vector of carefully crafted features. Second, using this dataset, we conduct a comparative analysis of different machine-learning techniques and features for classification. Finally, we develop two types ofclassifiers: based on the Boosted tree algorithm and the neural network-based classifier. Both are able to attribute a Bitcoin address to one of the predefined classes/companies. Our binary classification model achieves an F1 score of 76% using the Boosted tree algorithm, while our deep learning model achieves a 90% F1 score for multi-class classification with an accuracy of 92% and 28% higher than related work correspondingly. We achieve 67% accuracy for linking Bitcoin addresses to one of the 180 companies with our deep-learning model. Acknowledgments I am profoundly honored to express my heartfelt appreciation to my distinguished supervisor, Dr. Ivan Pustogarov, for his unwavering and invaluable support that has been instrumental in the successful completion of my master’s degree. Throughout this academic journey, I have had the privilege of benefiting from Dr. Pustogarov’s profound wisdom, profound knowledge, and exceptional mentorship. His remarkable dedication to my growth and development as a researcher and scholar has left an indelible mark on my academic and professional life. I am deeply grateful for his patience, unwavering motivation, and the wealth of expertise he has shared with me. Working under the close guidance of Dr. Pustogarov has been a truly enriching experience. His ability to inspire with innovative ideas, provide construct

FAQs

What problem does the thesis address for Bitcoin analysis?

It addresses Bitcoin address classification and clustering, aiming to attribute a Bitcoin address to predefined entity classes and to specific companies.

How are classes and companies defined in the proposed approach?

It uses five coarse-grained classes (exchanges, online marketplaces, mining pools, fundraising/charity platforms, and gambling) and 180 companies for fine-grained classification.

What datasets and models are proposed, and what performance is reported?

The thesis builds a labeled dataset of 3 million addresses (2016–2022) with crafted feature vectors, then compares traditional ML methods and proposes boosted-tree and neural deep-learning classifiers with reported F1/accuracy results for multi-class and fine-grained company linking.

Aug 4, 2026

Upload by Ezra