An Extensive Study on Financial Fraud Detection Using Artificial Intelligence-Based Models

In recent years, financial fraud—defined as the fraudulent gaining of financial assets—has develop a progressively pressing issue for organizations of all stripes. Traditional techniques of fraud detection, such as manual inspections and verifications, are costly, inefficient, and prone to error. With the arrival of machine learning technologies, made possible by advancements in artificial intelligence, it is now feasible to notice fraudulent transactions by intelligently assessing a massive sum of financial data. By evaluating and synthesizing the existing literature, this work hopes to suggest an SLR that achieves this for the field of ML-based fraud finding. For this review, we active the Kitchenham technique, which involves using a set of predetermined steps to collect, analyze, and report on the most relevant articles. Using the search criteria of prominent online library databases, many reports have been created. Based on the inclusion and exclusion standards, 93 papers were chosen, synthesized, and assessed. This article covers the most common type of fraud, as well as assessment metrics and a description of popular ML algorithms for fraud detection. The bulk of documented instances of credit card theft include the usage of ANNs and SVMs, two of the most prominent ML algorithms used for fraud detection.