SFIX:Scalable Financial-oriented Interpretable eXplanation

Abstract

The use of artificial intelligence in finance undoubtedly has a significant contribution in providing financial services to customers in a more efficient and secure manner. However, black box artificial intelligence algorithms can pose challenges in ensuring the safe functioning of financial services and monitoring desired outcomes. In this study, we have tried to design an explainable artificial intelligence method called Scalable Financial-oriented Interpretable eXplanation (SFIX) specific to the finance sector. While designing the SFIX method, time-based approaches such as fraud detection, credit scoring, customer profiling and other applications used in finance were taken into account. The accuracy and consistency of the dataset are also included in the calculations to support explainability. Finally, a simplified version of the SFIX method is also designed for quick testing of the model in case of problems in finding the real dataset.