Cyber Threats in Decentralized Finance (DeFi): Machine-Learning Approaches to Fraud and Protocol Attack Detection
Adaptive Defense and Intelligent Security Architectures for AI-Driven Cybersecurity, ArunKumar Sivaraman,Ajmery Sultana,Kamalavelu Velayutham,Janakiraman Nithiyanantham, Editör, IGI Global, New York, ss.365-400, 2026
- Yayın Türü: Kitapta Bölüm / Araştırma Kitabı
- Basım Tarihi: 2026
- Doi Numarası: 10.4018/979-8-2600-5116-0.ch012
- Yayınevi: IGI Global
- Basıldığı Şehir: New York
- Sayfa Sayıları: ss.365-400
- Editörler: ArunKumar Sivaraman,Ajmery Sultana,Kamalavelu Velayutham,Janakiraman Nithiyanantham, Editör
- Çanakkale Onsekiz Mart Üniversitesi Adresli: Evet
Özet
This chapter introduces a mechanism-to-signal approach to detecting fraud activities and protocol attacks in Decentralized Finance (DeFi) through machine learning. It maps to observables on-chain the mechanics of exploits in attacking the smart contract (e.g. smart contract exploits, economic manipulation, phishing, cross-chain bridge failure, transaction order abuse through transacts and erroneous call tracing) down to observable transactions, call traces, event logs, and state transitions. A structured taxonomy is the way to link these signatures to the supervised, anomaly-based, graph, and temporal models subject to such constraints as label scarcity, class imbalance, concept drift, and latency. Producible features of engineering, risk calibration and cost-wise evaluation are emphasized in this methodology. The deployment patterns combine the model outputs with governance and audit requirements, integrating the detection within an operational accountability architecture instead of a solution-oriented predictive model.