Privacy-Preserving Financial Transaction Pattern Recognition: A Differential Privacy Approach
Year: 2025
Type: article
Abstract: This document presents a new way to the precaution of business models recognized by privacy practices. This points out the main challenge of protecting data changes when managing the tasks in the acknowledgment. By implementing a multi-layered privacy protection architecture, the framework incorporates adaptive noise addition strategies and dynamic privacy budget allocation mechanisms specifically designed for financial transaction data. The only way to use a short period of memory (LSTM) with confidentiality (DP workshop is used by the specified number of funding Large Weighing There are 1.2 million files. Experimenting that the structure has completed ε = 1.0, while the privacy paid others are polite. The framework increases better matching for the confidentiality of privacy and has a 23.7% decrease in shock. The theoretical analysis proves that the framework provides formal ε-differential privacy guarantees while preserving essential transaction pattern features. This research makes the field of privacy-kept analysis data from the resolution of the defense of self-defense.
Authors: Zhonghao Wu, Zhengyi Zhang, Qiwen Zhao, Lei Yan
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W4408676406