FAIRNESS-AWARE GRAPH CONTRASTIVE LEARNING FOR FRAUD DETECTION IN FINANCIAL NETWORKS

Financial fraud detection has become increasingly critical as digital transactions proliferate across global financial networks. Traditional machine learning approaches often exhibit bias against certain demographic groups and fail to capture complex relational patterns inherent in financial transaction networks. This paper proposes a novel fairness-aware graph contrastive learning framework that simultaneously addresses algorithmic bias and improves fraud detection accuracy in financial networks. Our approach leverages graph neural networks (GNNs) enhanced with contrastive learning mechanisms while incorporating fairness constraints to ensure equitable treatment across different user groups. The framework introduces a dual-objective optimization strategy that balances fraud detection performance with fairness metrics, utilizing counterfactual graph augmentation techniques to mitigate discriminatory patterns. Experimental results on real-world financial datasets demonstrate that our method achieves superior fraud detection accuracy while significantly reducing bias compared to existing approaches. The proposed framework represents a significant advancement in developing trustworthy artificial intelligence systems for financial fraud detection that maintain both effectiveness and ethical considerations.