Real-Time Fraud Detection in Financial Transactions: Leveraging Machine Learning and Stream Processing for Dynamic Risk Assessment

EasyChair Preprint 14335

Abstract

In the rapidly evolving financial landscape, real-time fraud detection has become paramount for mitigating financial risks and safeguarding assets. This paper explores the integration of machine learning and stream processing technologies to enhance dynamic risk assessment in financial transactions. By leveraging advanced machine learning algorithms, such as anomaly detection and supervised learning models, in conjunction with stream processing frameworks, we present a novel approach to identifying and responding to fraudulent activities as they occur. Our methodology incorporates real-time data ingestion, processing, and analysis to detect suspicious patterns and anomalies with minimal latency. We evaluate the effectiveness of this approach using a comprehensive dataset of financial transactions, demonstrating significant improvements in detection accuracy and response times compared to traditional methods. The results highlight the potential of combining machine learning and stream processing to create a robust, adaptive fraud detection system that can dynamically assess and mitigate risks, offering a substantial advancement in the field of financial security.

Keyphrases: Leveraging, Recurrent Neural Networks (RNNs), machine learning

@booklet{EasyChair:14335,
  author    = {Abi Litty},
  title     = {Real-Time Fraud Detection in Financial Transactions: Leveraging Machine Learning and Stream Processing for Dynamic Risk Assessment},
  howpublished = {EasyChair Preprint 14335},
  year      = {EasyChair, 2024}}