Applying Machine Learning to Detect Fraud of Financial Statements: A Systematic Literature Review

The application of machine learning (ML) approaches to financial statement fraud detection is reviewed in this systematic literature review, highlighting the pressing need to improve fraud detection mechanisms in the financial markets. The primary objective of this study is to compile the body of current research, discover gaps in the field, and recommend future directions for research. Using the systematic literature review (SLR) approach, the study carefully reviews relevant primary literature to determine which algorithms for machine learning work best for fraud detection. The evaluation cautiously selects research from numerous databases, highlighting machine learning's promise and ability to detect fraud in financial statements. Findings show that complex machine learning techniques—in particular, deep learning methods and ensemble learning models—significantly improve the accuracy of fraudulent activity detection by recognizing intricate patterns and abnormalities that conventional techniques might miss. Despite these advancements, the review identifies a need for more inclusive datasets and the integration of non-financial data to augment the models' predictive power. The paper emphasizes the uniqueness and significance of utilizing machine learning (ML) in identifying financial fraud and advocates for additional research and development in integrating various data formats to enhance fraud detection models.