FedGAT-DCNN: Advanced Credit Card Fraud Detection Using Federated Learning, Graph Attention Networks, and Dilated Convolutions

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

In this paper, the authors propose a novel approach called FEDGAT-DCNN for credit card fraud detection. Their model integrates Graph Attention Networks (GAT) and dilated convolutions within a federated learning framework to address key challenges in fraud detection, such as limited data, rapidly evolving fraud techniques, and the need to detect subtle fraud patterns while maintaining data privacy. The authors demonstrate the effectiveness of their approach through experiments on the 2018CN and 2023EU datasets, where FEDGAT-DCNN outperforms traditional models and other federated learning methods.

 

(+) The paper addresses critical challenges in credit card fraud detection, particularly for small to medium-sized financial institutions with limited data resources.

(+) The integration of federated learning, GAT, and dilated convolutions is a novel and promising approach that enhances fraud detection accuracy, adaptability to new fraud patterns, and the ability to detect complex fraudulent activities while preserving data privacy.

(+) The authors provide a clear and detailed description of their methodology, including the construction of the transaction similarity graph, the application of dilated convolutions, and the integration of GAT within the federated learning framework.

(+) The experimental results on the 2018CN and 2023EU datasets demonstrate the superior performance of FEDGAT-DCNN compared to traditional models and other federated learning methods, highlighting its potential for real-world fraud detection scenarios.

 

(-) The paper could benefit from a more detailed discussion of the datasets used in the experiments, including their characteristics, size, and any preprocessing steps applied.

(-) While the authors mention the challenges of limited data and the need for continuous model updates, it would be helpful to provide more specific examples or case studies illustrating how FEDGAT-DCNN addresses these issues in practice.

(-) The paper could be strengthened by including a more in-depth comparison with other state-of-the-art methods in credit card fraud detection, discussing their relative strengths and weaknesses.

(-) An analysis of the computational complexity and scalability of FEDGAT-DCNN would be valuable to assess its feasibility for large-scale fraud detection systems.

Comments on the Quality of English Language

The quality of the English language is suitable for publication. The paper is well-written, with a clear and coherent structure. The authors demonstrate a good command of the English language, using appropriate technical terminology and maintaining a professional tone throughout the manuscript. 

Author Response

Comment 1: [The paper could benefit from a more detailed discussion of the datasets used in the experiments, including their characteristics, size, and any preprocessing steps applied.]

Response: Thank you for pointing this out. We agree with this comment. Therefore, we have rewritten the data processing part [Section 4.2, insert in line 486 ] to include additional information on the characteristics, size, and preprocessing steps applied to the datasets.

Modified Manuscript Text:

[Our dataset is randomly divided into multiple non-overlapping subsets based on the number of predetermined financial institutions, with each subset assigned to only one institution. This method ensures that the data between any two financial institutions is inconsistent, simulating the complete isolation of customer data across different institutions in reality. This careful division allows us to more accurately mimic real-world scenarios, thereby enhancing the credibility of our experimental results.

Once the data is split into multiple non-overlapping subsets, we further partition the subsets within each federated learning client. The data owned by each client is indepen-dently shuffled to ensure that the data partitioning of each financial institution is unique and representative of the wider dataset. Following standard experimental protocols, we divide each client’s dataset into a training set, validation set, and test set in a 7:2:1 ratio.

After partitioning the datasets for all federated learning clients, we construct a transaction similarity graph for each client. For any transaction record in the dataset, we calculate its similarity to the remaining transaction records and select the top-k most similar records as neighbors. This approach transforms the fraudulent transaction detection task into a semi-supervised node classification task on the graph.

This balanced approach is crucial for enabling models developed by various financial institutions to generalize effectively, ultimately leading to more effective and reliable fraud detection results. Our comprehensive strategy ensures that our federated learning framework achieves high performance while adhering to stringent security and privacy standards.]

Comment 2: [While the authors mention the challenges of limited data and the need for continuous model updates, it would be helpful to provide more specific examples or case studies illustrating how FEDGAT-DCNN addresses these issues in practice.]

Response: We agree with this suggestion and have included a specific example in the manuscript. [The revised text is highlighted below and has been inserted in line 78.]

Modified Manuscript Text:

[Small financial institutions in a developing country often struggle with limited access to comprehensive transaction data, making traditional fraud detection models less effective. By implementing FEDGAT-DCNN, these institutions can leverage collaborative training with larger banks via federated learning without compromising customer data privacy. Through this setup, even with sparse local data, the institutions continuously receive model updates informed by a wider network of transaction patterns. This ongoing refinement enhances the model's responsiveness to emerging fraud techniques, demonstrating a significant improvement in detection rates within just a few months of implementation.]

Comment 3: [The paper could be strengthened by including a more in-depth comparison with other state-of-the-art methods in credit card fraud detection, discussing their relative strengths and weaknesses.]

Response: Agree. We have expanded [Section 2.4 to include a more comprehensive comparative analysis with state-of-the-art methods in credit card fraud detection. ] The revised section is highlighted below:

Modified Manuscript Text:

[2.4. Comparative Analysis with State-of-the-Art Methods

Credit card fraud detection has evolved significantly with advancements in machine learning technologies. Traditional methods such as Support Vector Machines (SVM) and Random Forests are well-regarded for their robustness but often require extensive feature engineering and struggle with non-linear data. Furthermore, these approaches typically necessitate centralized data storage, raising significant privacy concerns.

Graph Neural Networks (GNNs), particularly Graph Attention Networks (GAT), have been adapted to better understand the complex relationships within transaction data networks. While GNNs offer improved pattern recognition over traditional algorithms, they usually require large amounts of data, which can be impractical in settings with strict data privacy regulations.

Federated learning offers a solution to the privacy issues by allowing models to be trained collaboratively without centralizing the data. However, this approach can suffer from issues related to data heterogeneity, where the diverse data distribution can lead to decreased model accuracy.

Our FEDGAT-DCNN model merges the strengths of GNNs with the privacy-preserving characteristics of federated learning. By integrating dilated convolutions into the GAT framework, FEDGAT-DCNN extends the model’s receptive field, enhancing its ability to detect subtle and complex fraudulent activities without additional computational costs. This integration not only maintains high accuracy in heterogeneous data environments but also ensures compliance with privacy standards, making it uniquely effective for real-world applications. Thus, FEDGAT-DCNN outperforms traditional models and standard federated learning approaches by providing a more adaptable, efficient, and privacy-aware solution in the landscape of credit card fraud detection.]

Comment 4: [An analysis of the computational complexity and scalability of FEDGAT-DCNN would be valuable to assess its feasibility for large-scale fraud detection systems.]

Response: Agree. We have supplemented the required time complexity analysis in [Section 3.5. ]The revised section is highlighted below: 

Modified Manuscript Text:

[3.5. Time Complexity Analysis

In this section, we thoroughly analyze the time complexity of the proposed method to assess its scalability on large datasets. The model comprises two main components: the dilated convolutional network and the GAT layer.

The dilated convolutional network module is responsible for computing dilated convolutional embeddings by aggregating information from the neighborhood of each node in the graph. The detailed time complexity analysis for this component is as follows:

• Neighbor Feature Extraction:For each node v the features of its d neighbors are extracted. This operation involves accessing the feature matrix and has a complexity of O(N ·d· F), where N is the number of nodes, d is the node degree, and F is the feature dimension.
• Dilated Convolutional Transformations:The neighbors'  features for each node are transformed through c dilated channels. For each channel, the node features are permuted and passed through a linear transformation, resulting in a complexity of O(N ·c·d·F²). The factor F² arises from the matrix multiplication involved in the linear transformation.
• Aggregation and Final Transformation:After applying dilated convolutions, the results are concatenated and transformed by an additional linear layer, contributing an additional O(N ·F²). complexity.
• Overall Complexity for the Dilated Convolution Layer: The overall complexity for the dilated convolution layer is:

This simplifies to O(N· c·d·F²). as the matrix multiplication and transformation are the dominant  operations.

The GAT layer processes the mixed features from the dilated convolutional network and applies attention mechanisms to dynamically weight the contributions of neighboring nodes. The specific time complexity analysis for the GAT layer is as follows:

• Attention Coefficient Calculation: For each edge, attention coefficients are computed using the node features, resulting in a complexity of O(E· F· nheads), where E is the number of edges and nheads is the number of attention heads.
• Feature Aggregation:Node features are aggregated based on the attention coefficients, leading to a complexity of O(N· F· nheads).
• Overall Complexity for the GAT Layer: The overall complexity for the GAT layer is:

The complexity is primarily dominated by the attention coefficient calculation, due to the potentially large number of edges in the graph.

The total time complexity of the model is determined by combining the complexities of the dilated convolutional network modules and the GAT layers. Considering the dominant terms from each component, the complexity can be expressed as:

The model's complexity depends on the number of nodes N, node degree d, feature dimension F, number of dilated channels c, number of edges E, and number of attention heads nheads. Optimizing these parameters is crucial for balancing model accuracy and computational efficiency, especially in large-scale graph data scenarios like credit card fraud detection.]

Reviewer 2 Report

Comments and Suggestions for Authors

Thank you for the "FedGAT-DCNN" work. My suggestions for minor revision are:

1. A summary of related work can be presented in tabular format.

2. Some of the references (above seven / ten years old) can be updated or removed.

3. The conclusion section (Section 5) can be presented in a point-wise fashion. 

Author Response

Comment 1:  [A summary of related work can be presented in tabular format.]

Response: We appreciate the suggestion and have included a comprehensive comparison of various methods in [Section 3.4.2], where we utilize a table to outline the distinct features, strengths, and weaknesses of each compared method.

Comment 2:Some of the references (above seven / ten years old) can be updated or removed.

Response: Thank you for your suggestion to update or remove older references. It is important to note that while some references in our manuscript, particularly those pertaining to traditional machine learning methods like SVM, Random Forests, and neural networks such as LSTM and CNN, are indeed over seven to ten years old, these methodologies remain foundational to the field of credit card fraud detection. Removing these references could disrupt the narrative and coherence of our paper, as these methods provide essential context for understanding the evolution and current state of fraud detection technologies. Therefore, [we have chosen to retain these references to ensure a comprehensive and informative discussion for readers who may be unfamiliar with the historical development of these techniques.] However, where possible, we have supplemented older references with more recent publications that reflect ongoing advancements and current trends in the field.

Comment 3: The conclusion section (Section 5) can be presented in a point-wise fashion.  

Response: Thank you for your valuable suggestion. We appreciate the feedback and have revised the conclusion section (Section 5) to present it in a point-wise fashion for better clarity and readability. [In this research, we have explored the effectiveness of the FEDGAT-DCNN model for detecting credit card fraud, yielding substantial insights and outcomes through comprehensive experimental validation and ablation studies. The key conclusions drawn from our research are:]

• [Robust Detection Performance: FEDGAT-DCNN significantly enhances fraud detection capabilities, particularly in scenarios with highly imbalanced data. It achieves an ROC-AUC of 0.9712 on the 2018CN dataset and 0.9992 on the 2023EU dataset, demonstrating robust performance across varied data distributions.
• Advanced Feature Integration: The integration of Graph Attention Networks (GAT) and dilated convolutions within the FEDGAT-DCNN framework allows for dynamic and efficient adaptation to emerging fraud patterns, enhancing the model's ability to capture complex transaction patterns and contextual information.
• Privacy and Collaboration: Utilizing federated learning, FEDGAT-DCNN enables multiple financial institutions to collaboratively train the model while preserving data privacy. This collaboration not only enhances the detection accuracy by leveraging diverse data sources but also ensures compliance with stringent data protection regulations.
• Future Research Directions: Going forward, we intend to refine the FEDGAT-DCNN model further by exploring additional graph-based techniques and expanding its applicability to other types of financial fraud. This future work aims to provide valuable guidelines for deploying effective fraud detection systems in real-world financial environments.]

Reviewer 3 Report

Comments and Suggestions for Authors

The paper discusses challenges in credit card fraud detection and proposes a graph attention-federated learning framework to address these challenges. The problem is interesting. Some comments that should be addressed are 

1. Notations are not always defined in the paper and should be corrected.

2. The number of references is low.

Overall, the paper is well written.

Comments on the Quality of English Language

Good

Author Response

Comment 1: [Notations are not always defined in the paper and should be corrected.]

Response: Thank you for your valuable suggestion. We have thoroughly checked and supplemented the formulas and symbols to ensure clarity and completeness. [The changes can be found in line 289 and line 294.]

Modified Manuscript Text:

Line 289: h_v  is the aggregation of the dilated embeddings from all dilated convolution channels.

Line 294: where  h^mix_v denotes the node features for node v after mixing multi-hop neighbor information.

Finally, the mixed features h^mix_v  ∈  H^mix are fed into the current GAT layer to compute the node embeddings for the layer l.

Comment 2: [The number of references is low.]

Response: Agree. Therefore, we have added several references to the manuscript based on your suggestions, and their specific locations are as follows:

1. [New references have been added in lines 25, 28, and 31 of the manuscript.
2. New references have been added after Convolutional Neural Networks (CNN) in line 35.
3. New references have been added after (RNN) in line 36.
4. New references have been added after (VAE) in line 36.]

The new references are listed below:

1. Syaufi, A., Zahra, A. F., & Gholi, F. M. I. (2023). Employing Forensic Techniques in Proving and Prosecuting Cross-border Cyber-financial Crimes. International Journal of Cyber Criminology, 17(1), 85-101.
2. Hassan, M., & Aziz, L. A. R. (2023). The role of artificial intelligence in modern banking: an exploration of AI-driven approaches for enhanced fraud prevention, risk management, and regulatory compliance. Reviews of Contemporary Business Analytics, 6(1), 110-132.
3. Khalid, A. R., Owoh, N., Uthmani, O., Ashawa, M., Osamor, J., & Adejoh, J. (2024). Enhancing credit card fraud detection: an ensemble machine learning approach. Big Data and Cognitive Computing, 8(1), 6.
4. Mathew, J. C., Nithya, B., Vishwanatha, C. R., Shetty, P., Priya, H., & Kavya, G. (2022). An analysis on fraud detection in credit card transactions using machine learning techniques. In 2022 Second International Conference on Artificial Intelligence and Smart Energy (ICAIS) (pp. 265-272). IEEE.
5. Chen, J. I., & Lai, K. L. (2021). Deep convolution neural network model for credit-card fraud detection and alert. Journal of Artificial Intelligence, 3(02), 101-112.
6. Karthika, J., & Senthilselvi, A. (2023). Smart credit card fraud detection system based on dilated convolutional neural network with sampling technique. Multimedia Tools and Applications, 82(20), 31691-31708. Springer.
7. Forough, J., & Momtazi, S. (2021). Ensemble of deep sequential models for credit card fraud detection. Applied Soft Computing, 99, 106883. Elsevier.
8. Mrvčela, L., & Kostanjčar, Z. (2024). Probabilistic Deep Learning Approach to Credit Card Fraud Detection. In 2024 47th MIPRO ICT and Electronics Convention (MIPRO) (pp. 181-186). IEEE.

Reviewer 4 Report

Comments and Suggestions for Authors

Good experimentation. Well written paper. Satisfactory results. 

Author Response

Dear Reviewer,

Thank you for taking the time to review my paper and providing your valuable feedback. I appreciate that you did not have any suggestions for revisions, which gives me confidence in my work.

Thank you again for your time and assistance.