Abstract:
In the contemporary mobile banking sector, fraud transactions are becoming a huge
concern day by day due to inadequate knowledge, susceptibility to phishing, and the
propensity of individuals unfamiliar with banking practices, in such cases the victim
disclosing the OTP(One Time Password) to the deceptive callers. We used machine
learning (ML), precisely based on speech recognition, to cut down fraud activities. For
the datasets both fraud calls and legitimate calls are employed in this study. Remarkable
algorithms such as K-Nearest Neighbors (KNN), Support Vector Machine (SVM),
Random Forest, Logistic Regression, and Decision Tree Algorithm are utilized for
optimal speech recognition. The Random Forest Algorithm delivers 99% accuracy to
contribute an efficacious speech recognition framework. This approach is proposed to
detect fraud callers in the realm of online mobile banking for a robust solution during a
routine transaction. Predicted speech recognition determines the nature of anomaly
detection delivering a potential way to minimize fraud in the expanded mobile banking
sector. Methodologies like fraud caller detection have shown promising results that can
reduce this rising threat accurately. This is complete research on Reliability and
usefulness of the project, by reducing the risk of fraud transaction and enhancing the
capabilities of anomaly detection during transaction. Moreover, this approach will give an
immediate solution to the ordinary people who are facing such deceptive activities in
their financial transaction.
