
BMO Financial Group, in collaboration with the Democritus University of Thrace (DUTH) and IBM, has demonstrated a quantum machine learning feature-selection workflow on IBM superconducting processors, including the IBM Quantum Nighthawk r2 system. Reviewed by IBM’s quantum research team, the experiment applies quantum game theory to identify high-value predictive variables in financial datasets for downstream client modeling, risk assessment, and personalization workloads.
Rather than treating feature selection as a conventional classical statistical search, the team adapted the Proxy Quantum Games framework developed by researchers at BMO, DUTH, and IBM. Within this formulation, each candidate variable in a dataset is modeled as a player in a quantum strategic game, receiving higher payoff scores when aligned with the target outcome variable and lower scores when uninformative or redundant. According to BMO, the quantum ranking approach achieved feature-selection quality comparable to classical baselines while requiring one to two orders of magnitude fewer algorithmic samples, though specific qubit counts, circuit depths, and wall-clock runtime comparisons have not yet been published.
The milestone builds on BMO’s February 2025 entry into the IBM Quantum Network and the launch of the BMO Institute for Applied Artificial Intelligence & Quantum, alongside ecosystem collaborations with the Chicago Quantum Exchange and Quantum Industry Canada. BMO noted that the work remains at the early demonstration stage as the bank evaluates scalable hybrid integration within governed financial workflows.
Read BMO’s announcement here, view the overview on BMO’s AI portal here, and access the underlying Proxy Quantum Games paper in Parallel Processing Letters here.
November 19, 2026