Fairness in Fraud Detection Using Autoencoders and Gaussian Mixture Models

Fraud detection is critical in the financial sector to protect against
unauthorized transactions. However, anomaly detection models can inad-
vertently favor or discriminate against certain demographic groups, raising
concerns about fairness. In this project, we use Autoencoders and Gaus-
sian Mixture Models (GMM) on the IEEE-CIS Fraud Detection dataset
to identify and analyze potential biases. By implementing techniques such
as balanced sampling and adversarial debiasing, we aim to reduce biases
while maintaining high detection accuracy, thereby creating a fairer and
more reliable fraud detection system.