Abstract
The objective of this talk is to present work related to the mathematical foundations of machine learning (ML) explainability with applications to finance. Explainability algorithms are crucial for ML modeling in financial institutions in the U.S., particularly those involved in extending credit. These algorithms include generating reason codes for risk models, evaluating fraud and anti-money-laundering (AML) reasons in fraud/AML detection models, interpreting large language models and manifold learning methods used for clustering, and assessing algorithmic fairness in ML. The talk will explore how these explainability methods contribute to more transparent and accountable financial decision-making. Additionally, we will discuss the complexity and stability analysis of some game-theoretic attribution methods and their application to credit decision-making and fair lending.
Brief Biography
Alexey Miroshnikov is a Senior Principal Research Scientist at Discover Financial Services, Emerging Capabilities Research Team. Previously, he held various academic positions: Assistant Adjunct Professor at UCLA Mathematics Department (2016-2019), Postdoctoral Research Associate in the Department of Biostatistics and Epidemiology at UMass Amherst (2015-2016), and Visiting Assistant Professor in the Department of Mathematics and Statistics at UMass Amherst (2012-2015). He received my PhD from the University of Maryland, College Park in 2012. During his Ph.D., he worked as a researcher at the Institute of Applied and Computational Mathematics at FORTH in Crete, Greece.