A high-efficiency variational quantum classifier for high-dimensional data

Variational quantum algorithms (VQAs) are most promising to show quantum advantages on noisy intermediate-scale quantum devices. Variational quantum classifiers (VQCs) are widely applied to classification tasks in the quantum domain. However, VQCs cannot show advantages in high-dimensional data. The large number of features necessitates the use of a significant number of qubits in VQCs. This results in long training time and increases training difficulty, ultimately leading to poor classification performance. In this paper, in order to enhance the ability of VQCs to handle high-dimensional data, a high-efficiency variational quantum classifier (HE-VQC) is proposed. Comparative Qiskit simulations of HE-VQC and four common VQCs were conducted on the UNSW-NB15 dataset. The simulation results show that HE-VQC significantly reduces training time while delivering superior classification performance.