Skip to content

Scientific-Exchange-Anomaly-and-Fraud-Detection

SNF Scientific Exchanges Project: Anomaly and fraud detection in blockchain networks using hybrid K-means/EVT methods

View on GitHub

Information

Property Value
Language Python
Stars 0
Forks 0
Watchers 0
Open Issues 0
License No License
Created 2026-01-21
Last Updated 2026-02-19
Last Push 2026-01-21
Contributors 1
Default Branch main
Visibility private

Reproducibility

This repository includes reproducibility tools:

  • Python requirements.txt

Research Keywords

means Clustering, BlockchainFraud DetectionNetworksAnomaly DetectionMachine LearningExtreme Value TheoryCryptocurrencyK

Status

  • Issues: Enabled
  • Wiki: Enabled
  • Pages: Disabled

README

Anomaly and Fraud Detection in Blockchain Networks

SNF Scientific Exchanges Project (IZSEZ0_211195)

Project Overview

This project studies anomaly and fraud detection from the perspective of blockchain-based networks. Blockchain networks are increasingly being implemented into healthcare, supply chain, and retail systems through smart contracts and identity management. However, the immutability property means fraudulent transactions cannot be reversed, making rapid detection of anomalies critical.

Research Focus: - Evolution of blockchain-based network graphs over time - Static anomaly detection methods for blockchain networks - Dynamic anomaly detection methods using extreme value theory

Principal Investigators

Researcher Institution Role
Prof. Dr. Joerg Osterrieder ZHAW School of Engineering, Switzerland / BFH Main Applicant
Prof. Dr. Stephen Chan American University of Sharjah, UAE Co-Investigator

Research Methodology

The proposed methodology comprises three main components:

  1. Network Graph Analysis: Comprehensive analysis of blockchain network evolution, investigating power law models and network statistics (clustering coefficient, cliques, independent sets)

  2. Static Anomaly Detection: Hybrid approach combining K-means clustering with Extreme Value Theory (EVT) using Generalized Extreme Value (GEV) distributions

  3. Dynamic Anomaly Detection: Real-time detection using data streams and Generalized Pareto Distribution (GPD) with peaks-over-threshold methods

Key Contributions

  • Novel hybrid approach combining statistical methods with machine learning for blockchain anomaly detection
  • Dynamic detection methods based on extreme value theory for real-time fraud identification
  • Comprehensive analysis of blockchain network graph evolution
  • Reduced false positive rates compared to single-method approaches

Project Outcomes

  • Published research papers in international journals
  • R software packages and dashboard implementations
  • Presentations at Stanford's "The Science of Blockchain Conference"
  • Training of Masters students in blockchain research
  • Foundation for MSCA Industrial Doctoral Network (4.5M EUR)

Repository Structure

.
|-- methodology/
|   |-- static_detection.md     # K-means + EVT hybrid approach
|   |-- dynamic_detection.md    # GEV/GPD extreme value methods
|-- results/
|   |-- publications.md         # List of publications
|-- code/
|   |-- anomaly_detection.py    # Example implementation
|   |-- network_analysis.py     # Graph analysis utilities
|-- docs/
    |-- project_info.md         # SNF grant details

Keywords

Blockchain Fraud Detection Networks Anomaly Detection Machine Learning Extreme Value Theory Cryptocurrency K-means Clustering

Funding

This project was funded by the Swiss National Science Foundation (SNF) through the Scientific Exchanges program.

  • Grant Number: IZSEZ0_211195
  • Program: Scientific Exchanges
  • Duration: January - February 2023
  • Host Institution: American University of Sharjah, UAE

Contact

  • Prof. Dr. Joerg Osterrieder: ZHAW Profile
  • Prof. Dr. Stephen Chan: schan@aus.edu

This research contributes to improving security in blockchain-based networks by providing more accurate and efficient methods for detecting anomalies and fraud.