QUBO-Based Calibration for Regression Trees
Iro René Kouarfate ; Maxime Dion ; Anne MacKay, Mathieu Pigeon · 2026
Read QUBO-Based Calibration for Regression Trees by Iro René Kouarfate ; Maxime Dion ; Anne MacKay, Mathieu Pigeon (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-06528.txt.
Description
Iro René Kouarfate, Maxime Dion, Anne MacKay, and Mathieu Pigeon reformulate regression tree building as a sequence of quadratic binary optimization problems, known as QUBO. Regression trees choose splits step by step, and choosing splits for categorical predictors is combinatorial, where standard methods rely on heuristics that can miss the best split.
Their formulation handles the fractional objective with Dinkelbach's algorithm and a class-based encoding, so each problem's size depends only on the number of categories, not the sample size. Trees built with modern QUBO solvers match standard CART accuracy while finding higher-quality categorical splits. The work also opens a path toward hybrid classical-quantum tree building.
Catalog record
| CREATOR | Iro René Kouarfate ; Maxime Dion ; Anne MacKay, Mathieu Pigeon |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 91 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.06528