Submitted Abstract
Machine Learning (ML) has revolutionized business processes in many areas. In the financial domain, various FinTech initiatives have been contemplating the use of ML to explore the huge amount of available data. However, domain specific constraints, such as regulation in the financial realm, challenge the use of systems which can be perceived as black box. Indeed, ML processes implement oftentimes-convoluted algorithms and generally-complex algebra, leading to systems that are unable to provide an answer to an essential practical question:”How did you come up with this solution?”ExLiFT (Explainable Machine Learning in FinTech) proposes to investigate novel approaches, in co-design between practitioners and academics, for addressing the need for explainable ML. The outcomes will enable our industry partner to build systems that are reliably usable internally by the human operators in daily activities, and that are further in line with the concerns of external auditors.