Deep Neural Networks and Hybrid Architectures for Credit Card Fraud Prevention

DOI

10.2991/978-94-6239-801-6_152How to use a DOI?

Keywords

CNNs; LSTM; GAN; Credit Card Fraud

Abstract

Credit card fraud is becoming increasingly serious in recent years for people at home and abroad. The three representative ways in this paper are: Deep Neural Networks for multi-feature integration; a hybrid Convolutional Neural Network - Long Short-Term Memory (CNN-LSTM) with attention mechanisms for spatiotemporal transaction modeling; and Generative Adversarial Networks for dealing with class imbalance through synthetic fraud data generation. Experimental comparison on benchmark datasets shows that hybrid models integrating CNNs, LSTMs and attention mechanisms have superior performance, with an accuracy exceeding 99% and a recall of 0.89 on standard benchmarks. Based on the above, although different parts of a single deep learning model can be used for fraud detection, more effective results are achieved by combining several such models in a hybrid system. The paper also points out the following problems: data imbalance, lack of model interpretability, requirements for real-time processing, and changes in fraudulent behaviour. The above results offer a system of reference for researchers and practitioners engaged in the construction of strong, high-performance credit card fraud detection systems in actual financial environments.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Jiayi Liu
PY  - 2026
DA  - 2026/10/23
TI  - Deep Neural Networks and Hybrid Architectures for Credit Card Fraud Prevention
BT  - Proceedings of the 2026 2nd International Conference on Financial Innovation and Marketing Management (FIMM 2026)
PB  - Atlantis Press
SP  - 1448
EP  - 1455
SN  - 2352-5428
UR  - https://doi.org/10.2991/978-94-6239-801-6_152
DO  - 10.2991/978-94-6239-801-6_152
ID  - Liu2026
ER  -