QUANTUM MACHINE LEARNING ON NEAR-TERM QUANTUM DEVICES - CURRENT STATE OF SUPERVISED AND UNSUPERVISED TECHNIQUES

Aug 6, 2026

Over the past decade, rapid progress in quantum hardware—speed, qubit count, and quantum volume—has enabled a rise in Quantum Machine Learning (QML) deployments on real near-term devices. This survey concentrates on supervised and unsupervised learning applications actually executed on quantum hardware for real-world scenarios. It analyzes key implementation constraints, including data encoding, variational ansatz design, error mitigation, and gradient-based methods, and compares resulting performance against classical approaches. Finally, it summarizes current bottlenecks and suggests future directions to address them.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages43

Reading Time1 h 48 min

AccessFree to download

Introduction

Quantum hardware progress and QML motivation

Supervised, unsupervised, and reinforcement learning paradigms

Real-world application domains: HEP, finance, and healthcare

arXiv :2307 .00908v3 [ quant-ph] 8 Jun 2024 QUANTUM MACHINE LEARNING ON NEAR-TERM QUANTUM DEVICES: CURRENT STATE OF SUPERVISED AND UNSUPERVISED TECHNIQUES FOR REAL-WORLD APPLICATIONS Yaswitha Gujju Dept. of Computer Science, The University of Tokyo [yaswitha-gujju@g.ecc.u-tokyo.ac.jp](yaswitha-gujju@g.ecc.u-tokyo.ac.jp) Atsushi Matsuo IBM Quantum, IBM Research-Tokyo [matsuoa@jp.ibm.com](matsuoa@jp.ibm.com) Rudy Raymond∗ Global Technology and Applied Research, J.P. Morgan Chase & Co. Dept. of Computer Science, The University of Tokyo Quantum Computing Center, Keio University [raymond.putra@jpmchase.com](raymond.putra@jpmchase.com) ABSTRACT The past decade has witnessed significant advancements in quantum hardware, encompassing improvements in speed, qubit quantity, and quantum volume—a metric defining the maximum size of a quantum circuit effectively implementable on near-term quantum devices. This progress has led to a surge in Quantum Machine Learning (QML) applications on real hardware, aiming to achieve quantum advantage over classical approaches. This survey focuses on selected supervised and unsupervised learning applications executed on quantum hardware, specifically tailored for real-world scenarios. The exploration includes a thorough analysis of current QML implementation limitations on quantum hardware, covering techniques like encoding, ansatz structure, error mitigation, and gradient methods to address these challenges. Furthermore, the survey evaluates the performance of QML implementations in comparison to classical counterparts. In conclusion, we discuss existing bottlenecks related to applying QML on real quantum devices and propose potential solutions to overcome these challenges in the future. Keywords Quantum Machine Learning · Real hardware · Variational circuits · Quantum Kernel methods · Data encoding · High Energy Physics · Healthcare · Finance · Bottlenecks 1 Introduction Machine Learning (ML) is ubiquitous, with applications spanning image recognition, healthcare diagnosis, text translation, anomaly detection, and physics. In parallel, near-term quantum devices have shown potential in addressing classically intractable problems, even with the challenges of noise and limited qubit connectivity [89, 88] . While quantum factoring algorithms, such as Shor’s, remain challenging, there have been notable successes, like the factorization of N = 15 using nuclear spins as quantum bits with room temperature liquid state nuclear magnetic resonance techniques [229] . The combination of quantum computing [329, 327] and machine learning, termed, Quantum Machine Learning (QML) [157, 148, 4] has become an active research area with great advancements being made in the last decade.Within QML, subdomains arise based on the data and algorithm types, whether classical or quantum. In this survey, we delve into different aspects of QML, specifically focusing on algorithms that leverage real quantum hardware, either in supervised or unsupervised contexts. In addition to the paradigms mentioned, reinforcement learning (RL) represents the third paradigm. Although not addressed in our current survey, we direct readers to [180, 182, 181] for a comprehensive overview of the literature on quantum reinforcement learning. ∗Part of this work was written while RR was with IBM Research – Tokyo. (a) Applications of QML. These are the different subproblems identified among the papers surveyed for real world domains that include High Energy Physics, Healthcare and Finance. (b) Outline of QML. The data can be inherently quantum or classical depending on the application. Consequently, we perform quantum state preparation for classical data. Based on the papers reviewed, we study different encoding techniques for classical data. We primarily focus on Variational Quantum Circuit and Kernel models and study the drawbacks and current challenges in the field. Figure 1: Overview of the paper. The p

FAQs

What does “quantum volume” mean in the context of near-term devices?

Quantum volume is a metric that captures the maximum size of a quantum circuit that can be effectively implemented on near-term quantum devices.

Which real-world application domains are emphasized for QML implementations on hardware?

The survey focuses on high energy physics, finance, and healthcare, discussing how QML is applied in these domains on real quantum systems.

What main challenges does the survey analyze for running QML on quantum hardware?

It covers limitations and techniques including data encoding, variational circuit (ansatz) structure, error mitigation, and gradient methods used to manage hardware noise and optimization difficulties.

Aug 6, 2026

Upload by Olivia Brown