محمدمهدي يادگار

In recent years, the emergence of new technologies and the revolution in this field have led to the widespread integration of the internet and technology into financial and business discussions. This increase has created opportunities for malicious attackers to exploit various methods of fraud, resulting in significant costs for these companies. Proper fraud detection allows investigators to take timely actions and prevent further fraud and financial losses. In recent years, the use of machine learning algorithms for detecting fraud in financial data has gained attention. One of the major challenges in the field of financial data is the imbalance of datasets, which makes it difficult for machine learning models to perform effectively. Various methods have been proposed to improve this imbalance, each with its own advantages and disadvantages. In this research, after reviewing the concepts and previous works done in anomaly detection in financial data, we propose a new method for over sampling in imbalanced financial data. This new method, by generating new samples in each cluster of data and using an ensemble learning model, can improve the results. In the eva‎luation and comparison of the proposed method, ACVAE-S, with other models, common eva‎luation metrics have been used. In these comparisons, the proposed method, on average, has led to an increase in the F-measure (by +2), accuracy (by +4), and recall (by +3) compared to baseline models. Additionally, compared to previous works in this field, it has resulted in an average increase in the F-measure (by +4) and recall (by +8).

كشف كلاهبرداري , داده‌هاي مالي , يادگيري‌ماشين , يادگيري گروهي , يادگيري گروهي , متوازن‌سازي داده

fraud detection , machine learning , financial data , credit card , ensemble learning , data balancing

Mohamad Mahdi Yadegar

Hossein Rahmani