Research on the Application of Machine Learning in Financial Anomaly Detection

Abstract

Financial anomaly detection is crucial for maintaining market order and protecting investor interests. This study explores the application of machine learning in financial anomaly detection. Using comparative analysis, the research contrasts traditional statistical methods with various types of machine learning algorithms in financial anomaly detection performance. The study covers supervised learning, unsupervised learning, and deep learning methods, analyzing their advantages and limitations in handling high-dimensional financial data and identifying complex fraud patterns. The research finds that ensemble learning methods perform exceptionally well in balancing detection accuracy and model interpretability. However, the study also highlights challenges in machine learning applications, such as sample imbalance and model generalization issues. To address these problems, the study introduces hybrid models that combine domain knowledge with data-driven approaches. Finally, the research discusses the potential of explainable artificial intelligence in enhancing the credibility of financial anomaly detection. This study provides new insights for improving the effectiveness and efficiency of financial anomaly detection.