Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

< Back to Article

Table 1.

Summary of the advantages and disadvantages of the relevant work on Internet fraud transaction detection.

More »

Table 1 Expand

Table 2.

Term definition.

More »

Table 2 Expand

Table 3.

Symbol table.

More »

Table 3 Expand

Fig 1.

The overall framework of the THG-OAFN model.

More »

Fig 1 Expand

Algorithm 1.

THG-OAFN: Time-aware heterogeneous graph oversampling and active fraud detection algorithm.

More »

Algorithm 1 Expand

Fig 2.

The integration of GRU and GNN.

More »

Fig 2 Expand

Algorithm 2.

The GRU-GNN integration and fusion process.

More »

Algorithm 2 Expand

Fig 3.

The structure of the heterogeneous graph-based oversampling model.

More »

Fig 3 Expand

Fig 4.

The technical process of SMOTE.

More »

Fig 4 Expand

Algorithm 3.

Pseudocode based on a heterogeneous graph oversampling model.

More »

Algorithm 3 Expand

Fig 5.

The structure of the heterogeneous graph fraud detection framework based on multilayer attention.

More »

Fig 5 Expand

Algorithm 4.

The pseudocode for multi-layer attention fusion of relationship-neighborhood-information.

More »

Algorithm 4 Expand

Table 4.

The summary of the statistical characteristics of datasets.

More »

Table 4 Expand

Table 5.

Model performance comparison after hyperparameter adjustment (Amazon dataset, 40% training ratio).

More »

Table 5 Expand

Table 6.

The hyperparameter grid search results of the THG-OAFN model (Amazon dataset).

More »

Table 6 Expand

Fig 6.

The influence of oversampling degree on model performance.

More »

Fig 6 Expand

Table 7.

The regularization control effects at an oversampling level of 1.2 (Amazon dataset).

More »

Table 7 Expand

Table 8.

Ablation experiment results of the attention heads (AUC/%).

More »

Table 8 Expand

Table 9.

Module-level ablation experiment results.

More »

Table 9 Expand

Fig 7.

Influence of different node embedding dimensions on model performance ((a) AUC; (b) Recall; (c) F1).

More »

Fig 7 Expand

Fig 8.

Effects of various training ratios on model performance ((a) AUC; (b) Recall; (c) F1).

More »

Fig 8 Expand

Table 10.

Performance comparison between THG-OAFN and the baseline models on the Amazon dataset (70% training ratio).

More »

Table 10 Expand

Fig 9.

Performance of different models on various datasets ((a) Amazon; (b) YelpChi; (c) Credit card fraud detection dataset).

More »

Fig 9 Expand

Fig 10.

Performance of diverse datasets on different models ((a) AUC; (b) Recall; (c) F1).

More »

Fig 10 Expand

Table 11.

Performance comparison between THG-OAFN and the specialized fraud detection model.

More »

Table 11 Expand

Fig 11.

Node embedding PCA dimension reduction coordinates (sampling 1000 nodes in the Amazon dataset).

More »

Fig 11 Expand

Fig 12.

Multi-layer attention weight distribution.

More »

Fig 12 Expand

Table 12.

Contribution of gradient backpropagation features (TOP5 features of the credit card fraud detection dataset).

More »

Table 12 Expand

Fig 13.

The average cosine similarity of the proposed method under different nodes ((a) Amazon; (b) YelpChi; (c) Credit card fraud detection dataset).

More »

Fig 13 Expand

Table 13.

Comparison of execution time (in seconds).

More »

Table 13 Expand

Table 14.

Cross-domain generalization experiment: Performance comparison between THG-OAFN and baseline models on financial datasets.

More »

Table 14 Expand

Table 15.

Performance comparison of insurance claim fraud detection (F1-score/%).

More »

Table 15 Expand