Anomaly Detection in High-Dimensional Bank Account Balances via Robust Methods

2025-11-14

AI Summary

Existing anomaly detection methods for high-dimensional bank account balance data suffer from low efficiency and poor robustness. Method: This paper proposes a class of robust statistical methods designed specifically for medium-to-high-dimensional (tens to hundreds of dimensions) financial time-series settings, featuring both high breakdown point and low computational complexity. By reformulating covariance estimation and depth-function computation paradigms, the approach significantly reduces the computational overhead of traditional methods such as Minimum Covariance Determinant (MCD). Contribution/Results: Evaluated on 2.6 million daily real-world account balance records, the method achieves a 12–18% improvement in detection accuracy over baselines, processes daily data within minutes, and demonstrates strong robustness against noise and data contamination. To our knowledge, this is the first work to achieve Pareto-optimal trade-offs between efficiency and robustness for high-dimensional robust anomaly detection in large-scale financial monitoring, providing a deployable statistical foundation for real-time risk control systems.

Technology Category

• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Machine Learning: Ensemble Methods
• Natural Language Processing: Safety and Robustness

Application Category

• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Security and Privacy: Large-scale security measurements
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

Detecting point anomalies in bank account balances is essential for financial institutions, as it enables the identification of potential fraud, operational issues, or other irregularities. Robust statistics is useful for flagging outliers and for providing estimates of the data distribution parameters that are not affected by contaminated observations. However, such a strategy is often less efficient and computationally expensive under high dimensional setting. In this paper, we propose and evaluate empirically several robust approaches that may be computationally efficient in medium and high dimensional datasets, with high breakdown points and low computational time. Our application deals with around 2.6 million daily records of anonymous users'bank account balances.

Problem

• Detecting anomalies in high-dimensional bank account balances
• Developing robust methods resistant to data contamination
• Improving computational efficiency for large-scale financial datasets

Innovation

• Robust statistical methods for anomaly detection
• High-dimensional bank account balance analysis
• Computationally efficient with high breakdown points