A0396
Title: A quantum and binary incremental learning procedure for categorical and ordinal classification
Authors: Yueh Lin - IESEG School of Management (France) [presenting]
Luis Fernando Perez Armas - IESEG School of Management (France)
Stefano Nasini - IESEG School of Management (France)
Martine Labbe - Free University of Bruxelles (Belgium)
Abstract: Mathematical programming approaches are explored for categorical and ordinal classification, leveraging decision trees and disjunctive normal form representations. The equivalence conditions of these formulations are analyzed, and an incremental learning procedure that solves a sequence of reduced problems is developed. Both linear (MILP) and quadratic (QUBO) formulations are examined, solving the latter using quantum adiabatic/variational techniques. The incremental learning approach strategically reduces problem size by iteratively refining feature selection, enabling efficient solutions while preserving classification accuracy. Computational experiments demonstrate that the method significantly accelerates convergence while achieving optimal (or near-optimal) classification accuracy across benchmark datasets. Moreover, the results reveal that quantum adiabatic/variational methods can outperform traditional branch-and-bound and feasibility pump algorithms for small to medium-sized instances. These findings lay the foundation for a unified, optimization-driven approach to statistical classification, bridging classical and quantum computational paradigms.
