Strategic Data Re-Uploads: A Pathway to Improved Quantum Classification Data Re-Uploading Strategies for Improved Quantum Classifier Performance

TL;DR

This paper introduces a data re-uploading strategy in quantum machine learning that enhances classifier accuracy and robustness, outperforming existing models across various datasets and optimization methods.

Contribution

It proposes a novel data re-uploading approach for quantum classifiers, analyzing its effectiveness with different cost functions, datasets, and optimization techniques.

Findings

Re-uploading classical data improves quantum classifier accuracy.

The approach outperforms existing models on multiple datasets.

Different cost functions and optimization methods impact performance.

Abstract

Quantum machine learning (QML) is a promising field that explores the applications of quantum computing to machine learning tasks. A significant hurdle in the advancement of quantum machine learning lies in the development of efficient and resilient quantum classifiers capable of accurately mapping input data to specific, discrete target outputs. In this paper, we propose a novel approach to improve quantum classifier performance by using a data re-uploading strategy. Re-uploading classical information into quantum states multiple times can enhance the accuracy of quantum classifiers. We investigate the effects of different cost functions, such as fidelity and trace distance, on the optimization process and the classification results. We demonstrate our approach to two classification patterns: a linear classification pattern (LCP) and a non-linear classification pattern (NLCP). We evaluate the efficacy of our approach by benchmarking it against four distinct optimization techniques: L-BFGS-B, COBYLA, Nelder-Mead, and SLSQP. Additionally, we study the different impacts of fixed datasets and random datasets. Our results show that our approach can achieve high classification accuracy and robustness and outperform the existing quantum classifier models.