Quantum classifiers for domain adaptation

TL;DR

This paper introduces two quantum domain adaptation classifiers that leverage quantum algorithms to achieve significant speedups over classical methods, enabling efficient transfer learning in machine learning tasks.

Contribution

It presents novel quantum implementations of domain adaptation classifiers, including a QBLAS-based method and a variational hybrid quantum-classical approach, demonstrating quantum speedup.

Findings

QBLAS-based classifier predicts labels with logarithmic resource complexity.

Hybrid quantum-classical classifier efficiently performs domain adaptation.

Quantum methods outperform classical counterparts in speed and resource usage.

Abstract

Transfer learning (TL), a crucial subfield of machine learning, aims to accomplish a task in the target domain with the acquired knowledge of the source domain. Specifically, effective domain adaptation (DA) facilitates the delivery of the TL task where all the data samples of the two domains are distributed in the same feature space. In this paper, two quantum implementations of the DA classifier are presented with quantum speedup compared with the classical DA classifier. One implementation, the quantum basic linear algebra subroutines (QBLAS)-based classifier, can predict the labels of the target domain data with logarithmic resources in the number and dimension of the given data. The other implementation efficiently accomplishes the DA task through a variational hybrid quantum-classical procedure.