Enhanced artificial intelligence-based cybersecurity for the detection of cyber fraud in the banking industry

Enhanced artificial intelligence-based cybersecurity for the detection of cyber fraud in the banking industry

The financial and banking industries have acknowledged the significance of creating credit card cyber fraud detection systems for a considerable period. Despite these efforts, businesses are still struggling with an increase in credit card cyber fraud. These incidents are fuelled by the ongoing technological revolution, which relies heavily on key enabling technologies such as Artificial Intelligence (AI)/Machine Learning (ML), big data, cloud computing, and the Internet of Things (IoT). Cybersecurity has become a paramount concern in the banking industry owing to the widespread occurrence of breaches and crimes. The growing prevalence of cybersecurity data breaches has undermined the efficacy of existing cyber fraud detection systems in detecting intricate criminal activities. The challenge of detecting credit card cyber fraud is exacerbated by two factors: the absence of enriched datasets and class imbalance. Credit card cyber fraud is a significant cybersecurity concern for the banking system worldwide, particularly as the number of financial transactions using credit cards increases. Diverse methodologies have been used to counteract these threats. Conventional anomaly detection is frequently employed; however, it tends to be time-consuming, resource-intensive, and imprecise. Artificial intelligence has the potential to significantly improve the accuracy of cyber fraud detection. It is imperative to promptly utilise experimental approaches that integrate machine learning and deep learning (DL) techniques to identify fraudulent activities and conduct factor analysis on anonymised credit card data. This will facilitate a more comprehensive comprehension of the interrelationships among various features and provide invaluable perspectives for counteracting credit card cybercrime. This research assesses several machine learning and deep learning techniques that are routinely employed to address credit card cyber fraud and binary cybersecurity problems. We created and evaluated three innovative cyber fraud detection models, each specifically designed to enhance the accuracy and efficiency of credit card transaction cyber fraud detection algorithms. These include a novel hybrid ML model, a novel hybrid DL model (CNN-BiLSTM), and a novel hybrid ML+DL model. The experiments demonstrate that the novel hybrid ML model, combined with the stacking ensemble, outperforms other individual ML models in detecting credit card cyber fraud. It also achieved the highest performance in credit card cyber fraud detection compared to other individual ML models. In addition, the novel hybrid DL model (CNN-BiLSTM) surpasses other individual DL models in cyber fraud detection. Furthermore, the novel hybrid ML+DL model with a stacking ensemble surpassed the hybrid ML model, hybrid DL model, and all individual models. This research further presents a theoretical structure based on empirical evidence from the actual world. The innovative ensemble hybrid ML + DL model represents the highest level of cyber fraud detection capabilities, leading to improved security in financial transactions. In summary, the results of the experiments clearly demonstrate that the proposed models were capable of producing accurate outcomes for the detection of cyber fraud in credit card systems and could be implemented in banking systems. As prominent artificial intelligence tools that enable the more accurate detection of cyber fraud, the newly developed models make significant contributions to the banking industry. As a result, these methods could be implemented to safeguard financial transactions through the implementation of a more coherent, accurate, and efficient methodology.

File reproduced in accordance with the copyright policy of the publisher/author/creator.