Enhancing the Explainability of Temporal Financial Data

Deep Learning has seen great success in recent years in application to high-dimensional data. In the advent of increased computability, the increase in accuracy of Deep Learning has surpassed traditional statistical methods. Consequently, the complexity of Deep Learning methods are not interpretable (black-box); therefore, are often dismissed in high-risk domains such as health care and finance. Explainable Artificial Intelligence (XAI) aims to increase the interpretability of black-box models through explanations. In finance, forecasting is important for both the researchers and policymakers in financial economics, but the development of XAI for time-series models is limited and often the XAI methods do not take into account the aspect of time in explanations. Therefore, we propose to develop and implement a variant of the deep learning explainer Integrated Gradients to produce explanations for temporal data.

Faculty Supervisor:

Hsuan Fu

Student:

Partner:

Swansea University

Discipline:

Computer science

Sector:

Education

University:

Université Laval

Program: