Secure Optimization of API-Driven Financial Transactions Using Deep Learning: A Threat Detection Framework for Mutual Fund Processing

Abstract

For software applications and systems to interact smoothly and support automated and efficient service delivery, system-to-system communication via Application Programming Interfaces (APIs) is crucial. APIs enable the sharing of data and functions across various platforms, improving both operational performance and user interaction. However, this integration can expose systems to security threats that may be exploited by malicious entities, emphasizing the need to recognize and address related security risks. In this paper, secure optimization of API-driven financial transactions using deep learning a threat detection framework for mutual fund processing (SO-APID-FT-DL-TDF-MFP) is proposed. At first, the input data is taken from the CIC-IDS2017 dataset. Then, the gathered data are fed into the pre-processing segment using implicit unscented particle filter (IUPF) which is used to eliminating noise. The pre-processed data are fed into Gegenbauer graph neural networks (GGNN) for prediction purpose. GGNN is used to predict potential security threats in the API-driven financial transactions by identifying irregular patterns and anomalies in the transaction data, thereby enhancing the overall security of the mutual fund processing system. Then, the proposed method implemented in python and the performance metrics like accuracy, precision, F1-score, recall, receiver operating characteristic (ROC) and specificity analyzed. The proposed SO-APID-FT-DL-TDF-MFP achieves 98% precision, 97% recall, 96% F1-score, 97.1% specificity, 97.5% accuracy, and 1.149 seconds computational time, with a high ROC of 0.99 compared with existing methods, such as adoption of deep-learning models for managing threat in API calls with transparency obligation practice for overall resilience (MT-APIC-TOP-OR-DL), deep learning for intelligent assessment of financial investment risk prediction (IA-FIRP-DL) and fraud prediction using machine learning: the case of investment advisors in canada  (FP-CIAC-ML).