This lecture is divided into two sections, both focusing on the topic of explainable artificial intelligence (XAI). The first section provides an overview of various interpretation methods utilized in machine learning, such as local and global interpretation methods, optimization-based approaches, counterfactual explanations, and recent advancements in glass-box accurate learning algorithms. The second section is dedicated to a rule discovery method. Following an introduction to the method, its application to a real prediction problem in a large bank will be demonstrated.

Ilker Birbil is a professor of AI & Optimization Techniques for Business & Society in University of Amsterdam, where he is the head of the Business Analytics section of the Amsterdam Business School. His research interests center around optimization methods in data science and decision making. Lately, he is working on explainable artificial intelligence, optimization for machine learning, and data privacy in operations research.