City Research Online

Abstract

Third-party providers pose a significant risk for financial institutions owing to the manner in which banks expose their API to the public. Such threats include inadequate authentication, malicious injections, unsecure key handling. Therefore, financial institutions must adopt a series of countermeasures to mitigate threats exposed by third-party providers, and anomaly detection is considered one such method. In this paper, we develop random forests and a linear kernel SVM to compare the accuracy of our models in predicting anomalous user access behaviour. A dataset that presents users' access behaviour as a numerical feature, including raw API call graphs, is utilised as a case study in this paper. Our novel approach of identifying a risk score and predicting it with a deep neural network showed a high degree of accuracy when the risk scores were developed as a multi-class classification problem

Publication Type:	Conference or Workshop Item (Paper)
Publisher Keywords:	Open banking API security, Open Banking API Anomaly, Open banking API risks
Subjects:	H Social Sciences > HF Commerce Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Departments:	School of Science & Technology School of Science & Technology > Computer Science School of Science & Technology > Engineering
SWORD Depositor:	Symplectic Administrator

Export

Downloads