Prof. Dr. Samir Elhedhli
University of Waterloo, Canada
Samir Elhedhli is a Professor in the Department of Management Science and Engineering at the University of Waterloo, specializing in Data Analytics and Optimization. He has held research grants from NSERC, SSHRC, CFI, OCE, and MITACS, and has collaborated with industry partners in manufacturing, robotics, healthcare, logistics, supply chain management, and airline operations. His work has appeared in scientific journals such as Management Science, Mathematical Programming, Manufacturing and Service Operations Management, INFORMS Journal on Computing, IISE Transactions, Computers and Industrial Engineering and the European Journal of Operational Research, among others. Professor Elhedhli is a former President of the Canadian Operational Research Society, former Editor-in-Chief of its flagship journal, Information Systems and Operational Research, and former Chair of the University of Waterloo’s Department of Management Sciences.
Keynote
- Date
- Time
- RoomKhalifa University Auditorium
Optimization-Based Learning for Interpretable and Human-Inspired Binary Classification
We introduce a novel inverse optimization-based classifier for multi-attribute decision-making and binary classification problems. We develop interpretable classification rules that are inspired by human decision-making heuristics and propose direct and noisy inverse optimization frameworks for learning the underlying attributes from observed classifications. We establish the foundations of the classifier based on concepts from inverse optimization and interior-point methods, and benchmark it against several established machine learning methods, including random forests, support vector machines, logistic regression, ridge classifiers, decision trees, and neural networks. When evaluated on venture commercialization and heart disease datasets, the proposed classifier exhibits strong out-of-sample performance, demonstrating robust generalization and limited overfitting, and outperforms benchmark machine learning classifiers while retaining a transparent and interpretable decision structure.