With the advent of digital technology and the prevalence of online transactions, there has been a surge in different forms of fraud, particularly within the financial sector. Internet loans, while providing a convenient means for individuals to obtain rapid financial aid, have unfortunately become a prime target for deceitful schemes. Conventional fraud detection systems commonly depend on rule-based methods and statistical models. Rule-based systems employ pre-established rules to identify transactions that exhibit certain patterns indicative of fraudulent activity. Statistical models, such as logistic regression, analyze historical transaction data to identify anomalies. Although these methods have proven to be valuable, they frequently encounter difficulties in identifying intricate, non-linear patterns that are indicative of fraudulent activity in online loan applications.