Dynamic Anomaly Identification in Accounting Transactions via Multi-Head Self-Attention Networks

2025-11-15

AI Summary

To address the challenges of dynamically identifying covert anomalous behaviors in accounting transactions under stringent real-time requirements, this paper proposes a Transformer-based real-time anomaly detection method. The approach models transaction sequences as time-series matrices and employs multi-perspective feature aggregation, integrating positional encoding, multi-head self-attention, and regularization to effectively capture long-range dependencies and complex temporal patterns—enhancing model robustness and expressive power in low-dimensional representations. Extensive experiments on multiple public datasets demonstrate that our method consistently outperforms state-of-the-art baselines across key metrics, including AUC, F1-score, precision, and recall. Moreover, it exhibits strong stability under data perturbations and environmental shifts. This work delivers a practical, deployable solution for accounting risk control in high-throughput, high-complexity operational environments.

Technology Category

• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Machine Learning: Time-Series/Data Streams
• Natural Language Processing: Safety and Robustness

Application Category

• Responsible Web: Algorithmic accountability and transparency on the web
• Security and Privacy: Data transparency and provenance
• User Modeling, Personalization and Recommendation: Accountability, Transparency, and Ethics for personalization

Abstract

This study addresses the problem of dynamic anomaly detection in accounting transactions and proposes a real-time detection method based on a Transformer to tackle the challenges of hidden abnormal behaviors and high timeliness requirements in complex trading environments. The approach first models accounting transaction data by representing multi-dimensional records as time-series matrices and uses embedding layers and positional encoding to achieve low-dimensional mapping of inputs. A sequence modeling structure with multi-head self-attention is then constructed to capture global dependencies and aggregate features from multiple perspectives, thereby enhancing the ability to detect abnormal patterns. The network further integrates feed-forward layers and regularization strategies to achieve deep feature representation and accurate anomaly probability estimation. To validate the effectiveness of the method, extensive experiments were conducted on a public dataset, including comparative analysis, hyperparameter sensitivity tests, environmental sensitivity tests, and data sensitivity tests. Results show that the proposed method outperforms baseline models in AUC, F1-Score, Precision, and Recall, and maintains stable performance under different environmental conditions and data perturbations. These findings confirm the applicability and advantages of the Transformer-based framework for dynamic anomaly detection in accounting transactions and provide methodological support for intelligent financial risk control and auditing.

Problem

• Detects dynamic anomalies in accounting transactions in real-time
• Handles hidden abnormal behaviors in complex trading environments
• Uses multi-head self-attention to capture global transaction dependencies

Innovation

• Transformer-based real-time anomaly detection method
• Multi-head self-attention captures global transaction dependencies
• Embedding layers with positional encoding model time-series data