Parallelized variational quantum classifier with shallow QRAM circuit

In this paper, we present a novel quantum variational circuit, harnessing the capabilities of a pre-determined quantum random access memory (QRAM) circuit to enhance machine learning tasks. Our approach enables parallel training on entire datasets, facilitated by QRAM’s streamlined data loading onto qubits through probability amplitudes. This leads to a logarithmic reduction in the qubit width (number of qubits) concerning data dimension and dataset size. Our model is adaptable to various loss functions, rendering it suitable for binary and multi-class classification tasks. To validate our methodology, we conducted numerical simulations using well-established benchmark datasets, Iris and handwritten digits, with the Pennylane platform. Impressively, our approach yielded high classification accuracy in these illustrations. This work demonstrates promising potential for quantum machine learning applications, especially when dealing with large datasets.