Speaker
Description
Quantum kernel methods are among the most promising near-term applications of quantum machine learning, with the quantum feature map controlling a model's ability to capture meaningful correlations from little labelled data. We present a systematic approach that aligns a quantum-inspired feature map to highly imbalanced and scarce clinical datasets through kernel--target alignment (KTA) optimization, and feeds the resulting Gram matrix into a classical support vector machine (SVM). On six clinical datasets of therapeutic and prognostic relevance that are effectively intractable for a classical radial basis function (RBF) kernel SVM, the aligned quantum kernel improves the AUC by $\approx 0.22$ on average, while matching the classical baseline on a balanced benchmark. This suggests a practical NISQ-era pathway to analyze complex clinical data that is otherwise resistant to conventional machine learning.
| Topic | Quantum kernel methods and their applications |
|---|