CaT-GNN: Enhancing Credit Card Fraud Detection via Causal Temporal Graph Neural Networks

TL;DR

A novel method for credit card fraud detection, the CaT-GNN, which leverages causal invariant learning to reveal inherent correlations within transaction data and applies a causal mixup strategy to enhance the model's robustness and interpretability.

Abstract

Credit card fraud poses a significant threat to the economy. While Graph Neural Network (GNN)-based fraud detection methods perform well, they often overlook the causal effect of a node's local structure on predictions. This paper introduces a novel method for credit card fraud detection, the \textbf{\underline{Ca}}usal \textbf{\underline{T}}emporal \textbf{\underline{G}}raph \textbf{\underline{N}}eural \textbf{N}etwork (CaT-GNN), which leverages causal invariant learning to reveal inherent correlations within transaction data. By decomposing the problem into discovery and intervention phases, CaT-GNN identifies causal nodes within the transaction graph and applies a causal mixup strategy to enhance the model's robustness and interpretability. CaT-GNN consists of two key components: Causal-Inspector and Causal-Intervener. The Causal-Inspector utilizes attention weights in the temporal attention mechanism to identify causal and environment nodes without introducing additional parameters. Subsequently, the Causal-Intervener performs a causal mixup enhancement on environment nodes based on the set of nodes. Evaluated on three datasets, including a private financial dataset and two public datasets, CaT-GNN demonstrates superior performance over existing state-of-the-art methods. Our findings highlight the potential of integrating causal reasoning with graph neural networks to improve fraud detection capabilities in financial transactions.

Figures (5)

• Figure 1: The model overview. First Stage (discovery): we utilize an attention map in the attention temporal network to identify causal nodes and environment nodes. Second Stage: Intervention, we apply causal mix-up enhancement to the environment nodes.
• Figure 2: Motivation. The original prediction incorrectly identifies a fraudster (central node labeled $x_i$) as benign, as does the state-of-the-art GTAN model. Following our causal intervention, the prediction is correctly adjusted to identify $x_i$ as a fraudster. Green: benign users, red: fraudsters, gray: unlabeled nodes.
• Figure 3: The depiction of the proposed model's architecture, featuring a causal temporal graph attention mechanism, alongside the theoretical support for backdoor adjustment.
• Figure 4: The ablation study results on three datasets. Gray bars represent the D-CaT variant, blue bars represent the N-CaT variant, and orange bars represent the CaT-GNN model.
• Figure 5: Sensitivity analysis with respect to different training ratios (Left) and environment ratios (Right).