Theoretical and Applied Data Science Seminar - Managing Machine Learning Risk: Interpretability and Robustness, Dr. Agus Sudijianto

Event
Wednesday, April 14, 2021 - 4:25pm to 5:25pm
Event Type: 

Agus Sudjianto headshotTitle: Managing Machine Learning Risk: Interpretability and Robustness

LocationZoom

Abstract:  All models are wrong and when they are wrong they create financial or non-financial harm. Understanding, testing and managing potential model failures and their unintended consequences is the key focus of model risk management. This is a challenging task for complex machine learning models. Key critical enablers to anticipate and manage model failures include model interpretability and robustness. Despite the progress that has been made in explainable machine learning, post-hoc explainers are still fraught with weakness and complexity. In this talk, I will argue that what we need is a self-explanatory--inherently interpretable--machine learning model. I will discuss how to make sophisticated machine learning models such as Neural networks (Deep Learning) into self-explanatory models. This self-explanatory model construct also facilitates model robustness test and design, a critical aspect when models operate under dynamically changing or adversarial environment.

Bio: Agus Sudjianto is an executive vice president, head of Model Risk and a member of Management Committee at Wells Fargo, where he is responsible for enterprise model risk management.  Prior to his current position, Agus was the modeling and analytics director and chief model risk officer at Lloyds Banking Group in the United Kingdom. Before joining Lloyds, he was an executive and head of Quantitative Risk at Bank of America.  Prior to his career in banking, he was a product design manager in the Powertrain Division of Ford Motor Company.  Agus holds several U.S. patents in both finance and engineering. He has published numerous technical papers and is a co-author of Design and Modeling for Computer Experiments. His technical expertise and interests include quantitative risk, particularly credit risk modeling, machine learning and computational statistics.  He holds masters and doctorate degrees in engineering and management from Wayne State University and the Massachusetts Institute of Technology.

Category: