Intelligent Algorithms for the Detection of Suspicious Transactions in Payment Data Management Systems Based on LSTM...

Area of Science:

• Information Technology
• Artificial Intelligence
• Data Science

Background:

• Global efforts focus on advancing data processing and AI applications across various sectors.
• Optimizing socio-economic systems and ensuring database reliability in digital payment systems are critical.
• Increasing information reliability in payment systems is a significant challenge.

Purpose of the Study:

• To investigate methods for enhancing information reliability in payment information systems.
• To develop intelligent systems for detecting ambiguous suspicious transactions.
• To compare the performance of traditional and neural network models for fraud detection.

Main Methods:

• Analysis of ambiguous suspicious transaction characteristics in payment systems.
• Preliminary data preparation for intelligent detection of suspicious transactions.
• Development and comparative analysis of machine learning models, including traditional and neural network approaches.
• Integration of the Artificial Bee Colony (ABC) optimization algorithm with LSTM models for hyperparameter tuning.

Main Results:

• Established characteristics of ambiguous suspicious transactions.
• Developed preliminary data preparation stages for intelligent detection.
• Created and compared traditional and neural network machine learning models for detecting suspicious transactions.
• Optimized LSTM model performance using the ABC algorithm, leading to improved accuracy and efficiency in identifying complex fraudulent patterns.

Conclusions:

• The integration of the ABC algorithm significantly enhances the performance of LSTM models for suspicious transaction detection.
• Intelligent detection systems, optimized through advanced algorithms, are crucial for improving the reliability of payment information.
• This research contributes to more robust and efficient fraud detection in digital payment systems.