Year: 2025
Type: article
Abstract: Identification of bank cards fraud remains a serious worry for financial companies because of how sophisticated criminal behavior is becoming.. This study addresses this challenge by leveraging models increase the efficiency of detection systems. Utilizing a dataset from Kaggle, we implement and compare five distinct algorithms: LSTM networks, neural networks based on CNN, Decision Trees, Random Forests, and a Stacking Classifier. CNNs are employed to capture intricate patterns in transaction data, while LSTM address sequential dependencies and temporal dynamics. Decision Trees and Random Forests provide robust classification through hierarchical decision-making and ensemble learning. Additionally, a Stacking Classifier integrates the strengths of these algorithms to potentially improve overall performance. The comparative analysis of these methods tries to determine the best method for real-time identification of fraud. The results are expected to contribute significantly to the development of more secure credit card transaction systems, thereby mitigating financial losses and enhancing consumer trust.
Authors: P. Veeresh P. Veeresh, M. Thaher Basha M. Thaher Basha, Vajrala Varshith Vajrala Varshith, Beldar Taher Basha Beldar Taher Basha, Kadapala Giribabu Kadapala Giribabu
Institution: Jawaharlal Nehru Technological University Anantapur
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4410271854