A new approach to rating scale definition with quantum-inspired optimization
AI Breakdown
Get a structured breakdown of this paper — what it's about, the core idea, and key takeaways for the field.
Abstract
In finance, assessing the creditworthiness of loan applicants requires lenders to cluster borrowers using rating scales. Financial institutions must define the scales in compliance with strict institutional constraints, resulting in solving a complex combinatorial constrained optimization problem. This contribution studies how to solve this problem using a Quadratic Unconstrained Binary Optimization (QUBO) model, a formulation suitable for quantum hardware. We validate this approach by testing the proposed formulation with classical heuristics. We then benchmark the results against a brute-force method to demonstrate consistent solution quality and highlight the framework's suitability for more complex scenarios.