Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{176998,
        author = {Shoyeb Akhtar and Syed Jaid Mohd Ragib and Mohd Aslam Khan and Jameel Ur Rahman Khan and Faizan Ahmad},
        title = {AI Based Financial Fraud Detection In Banking System},
        journal = {International Journal of Innovative Research in Technology},
        year = {2025},
        volume = {11},
        number = {12},
        pages = {260-268},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=176998},
        abstract = {The digitalization of the financial sector introduces even more sophisticated forms of fraud to which traditional older detection mechanisms are not well adapted. This piece presents a detailed evaluation of machine learning models to detect fraud against several performance criteria. We implement five ML algorithms on a dataset of 284,807 transactions by European cardholders, such as Random Forest (RF), Decision Trees (DT), Logistic Regression (LR), K-Nearest Neighbour (KNN), and Naïve Bayes (NB). Our strategy includes Z-score normalization and then applying SMOTE to address class imbalance, effective feature selection, and class imbalance reduction. The optimal performance was observed with the RF model, which attained 99% accuracy, 99% precision, 98% recall, and 98% F1-score, with DT being not much behind. We also present comprehensive analyses of the inherent computational expense of the techniques, with RF having processed 1,000 transactions per second on off-the-shelf hardware. Practical implementation challenges are also addressed, including demands on latency (sub-200ms for fraud verification) and the necessity for explainable models. We conclude that ensemble models, especially Random Forests, VF reconciling accuracy (99%), computational time (0.8ms per prediction), and explanation for banking systems. We develop an innovative framework that harmoniously integrates these models into banking systems and alleviates compliance and explainability problems.},
        keywords = {Detection and Prevention of Fraud, Machine Learning, Security in Banking, Logistic Regression, Random Forest, Real-time Systems},
        month = {April},
        }