Scholarlink

In the era of increasing digital transactions and data exchange, safeguarding financial assets and data from fraudulent activities is a critical concern. Although existing algorithms have shown success, they face challenges in achieving a commendable level of precision for encrypted data. This research investigates the integration of Fully Homomorphic Encryption (FHE) into Logistic Regression and XGBoost models for credit card fraud detection, addressing a gap in existing literature. The results indicate that FHE does not significantly compromise model accuracy, with Logistic Regression and XGBoost achieving comparable performance on encrypted and plain text data. Real-world applicability is demonstrated using the ULB and Vesta datasets. The findings contribute to the discourse on privacy-preserving machine learning and offer insights for secure credit card fraud detection.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Fraud Detection Using Machine Learning Models and Encryption Techniques

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Jigyasa Agrawal,
  • Hien Nguyen

摘要

In the era of increasing digital transactions and data exchange, safeguarding financial assets and data from fraudulent activities is a critical concern. Although existing algorithms have shown success, they face challenges in achieving a commendable level of precision for encrypted data. This research investigates the integration of Fully Homomorphic Encryption (FHE) into Logistic Regression and XGBoost models for credit card fraud detection, addressing a gap in existing literature. The results indicate that FHE does not significantly compromise model accuracy, with Logistic Regression and XGBoost achieving comparable performance on encrypted and plain text data. Real-world applicability is demonstrated using the ULB and Vesta datasets. The findings contribute to the discourse on privacy-preserving machine learning and offer insights for secure credit card fraud detection.