Conservative quantum offline model-based optimization

TL;DR

This paper introduces COM-QEL, a hybrid quantum-classical approach that combines quantum extremal learning with conservative modeling to improve offline optimization reliability.

Contribution

It proposes integrating conservative objective models with quantum extremal learning to enhance generalization and avoid overly optimistic solutions in quantum offline optimization.

Findings

COM-QEL finds higher true objective solutions than original QEL.

The hybrid approach improves reliability in benchmark optimization tasks.

Conservative modeling safeguards against overestimation in unexplored regions.

Abstract

Offline model-based optimization (MBO) refers to the task of optimizing a black-box objective function using only a fixed set of prior input-output data, without any active experimentation. Recent work has introduced quantum extremal learning (QEL), which leverages the expressive power of variational quantum circuits to learn accurate surrogate functions by training on a few data points. However, as widely studied in the classical machine learning literature, predictive models may incorrectly extrapolate objective values in unexplored regions, leading to the selection of overly optimistic solutions. In this paper, we propose integrating QEL with conservative objective models (COM) - a regularization technique aimed at ensuring cautious predictions on out-of-distribution inputs. The resulting hybrid algorithm, COM-QEL, builds on the expressive power of quantum neural networks while safeguarding generalization via conservative modeling. Empirical results on benchmark optimization tasks demonstrate that COM-QEL reliably finds solutions with higher true objective values compared to the original QEL, validating its superiority for offline design problems.