The expansion of the e-commerce market has encouraged digital security companies to rethink card fraud detection methods. In a dynamic and constantly evolving environment, new fraud strategies are constantly created.
Neural networks encounter previously unseen statistical properties and may generate overconfident erroneous predictions. In order to specify model reliability and support interpretability, it is necessary to assess the uncertainty of the generated predictions.
A recent study on arXiv.org explores three different uncertainty qualification methods for uncertainty estimation of the transaction data available to the public. The interval between the probabilities extracted by the model and true probabilities is estimated. The researchers find out which method is most effective to capture prediction uncertainties.
Countless research works of deep neural networks (DNNs) in the task of credit card fraud detection have focused on improving the accuracy of point predictions and mitigating unwanted biases by building different network architectures or learning models. Quantifying uncertainty accompanied by point estimation is essential because it mitigates model unfairness and permits practitioners to develop trustworthy systems which abstain from suboptimal decisions due to low confidence. Explicitly, assessing uncertainties associated with DNNs predictions is critical in real-world card fraud detection settings for characteristic reasons, including (a) fraudsters constantly change their strategies, and accordingly, DNNs encounter observations that are not generated by the same process as the training distribution, (b) owing to the time-consuming process, very few transactions are timely checked by professional experts to update DNNs. Therefore, this study proposes three uncertainty quantification (UQ) techniques named Monte Carlo dropout, ensemble, and ensemble Monte Carlo dropout for card fraud detection applied on transaction data. Moreover, to evaluate the predictive uncertainty estimates, UQ confusion matrix and several performance metrics are utilized. Through experimental results, we show that the ensemble is more effective in capturing uncertainty corresponding to generated predictions. Additionally, we demonstrate that the proposed UQ methods provide extra insight to the point predictions, leading to elevate the fraud prevention process.
Research paper: Habibpour, M., “Uncertainty-Aware Credit Card Fraud Detection Using Deep Learning”, 2021. Link: https://arxiv.org/abs/2107.13508