Speaker
Description
Unsupervised feature selection is essential for high-dimensional machine learning, improving model quality and efficiency while providing interpretable insights. Existing methods struggle to simultaneously achieve robustness, interpretability, and computational efficiency, and feature selection remains largely unexplored in quantum machine learning despite the potential of quantum models to capture complex feature interactions. We propose the Quantum Hybrid Feature Selector (QHFS), an unsupervised feature selection framework combining quantum-enhanced feature extraction with interpretable scoring algorithms. Our contributions are threefold: (i) a quantum hybrid autoencoder pipeline coupling feature extraction with feature ranking in the original input space; (ii) three scoring mechanisms - SHAP-based scoring, correlation-matrix scoring (SCM), and weight-based analysis - relating latent features to original features without external predictive models; (iii) comprehensive empirical evaluation on synthetic and real-world benchmarks with statistical validation. On Madelon-style synthetic datasets, quantum SCM achieves up to 23.5% improvement in Mean Informative Rank and 9.7% in Informative Ratio over classical alternatives, with all improvements statistically significant at p<0.001. On the Communities and Crime dataset with injected noise features, both quantum and classical SCM variants achieve near-perfect noise elimination, demonstrating effective denoising on real-world data. We show that expert input is essential for evaluating feature selectors, as different methods emphasize distinct aspects of the data despite similar aggregate metrics. Finally, we evaluate QHFS under realistic hardware noise using device-informed simulations of IBM superconducting processors, demonstrating graceful degradation and systematic improvement with increased shot budgets, supporting practical viability on near-term devices. Published in EPJ Quantum Technology 13, 49 (2026).
| Topic | Hybrid quantum-classical approaches for learning and optimization |
|---|---|
| Additional Topic | Quantum algorithms for machine learning applications |