Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Predominant Aspects on Security for Quantum Machine Learning: Literature Review

Nicola Franco, Alona Sakhnenko, Leon Stolpmann, Daniel Thuerck, Fabian Petsch, Annika Rüll, Jeanette Miriam Lorenz

TL;DR

The paper addresses security in Quantum Machine Learning by performing a systematic literature review to map vulnerabilities and defenses in QML. It frames PQCs operating in a Hilbert space of dimension $2^n$ and analyzes how data encoding and hardware noise influence security, identifying non-classical attack vectors and upholding rigorous verification needs. Key contributions include a taxonomy of quantum-specific vulnerabilities (e.g., fault injections, cross-talk, Shuttles in ion traps) and defenses such as adversarial training, quantum differential privacy, MILP-based verification, and Lipschitz-based robustness measures, along with calls for cross-disciplinary benchmarking. The work offers guidance for researchers and practitioners seeking secure deployment of QML in real-world tasks and highlights the necessity of ongoing, experimentally validated security research in this rapidly evolving field.

Abstract

Quantum Machine Learning (QML) has emerged as a promising intersection of quantum computing and classical machine learning, anticipated to drive breakthroughs in computational tasks. This paper discusses the question which security concerns and strengths are connected to QML by means of a systematic literature review. We categorize and review the security of QML models, their vulnerabilities inherent to quantum architectures, and the mitigation strategies proposed. The survey reveals that while QML possesses unique strengths, it also introduces novel attack vectors not seen in classical systems. We point out specific risks, such as cross-talk in superconducting systems and forced repeated shuttle operations in ion-trap systems, which threaten QML's reliability. However, approaches like adversarial training, quantum noise exploitation, and quantum differential privacy have shown potential in enhancing QML robustness. Our review discuss the need for continued and rigorous research to ensure the secure deployment of QML in real-world applications. This work serves as a foundational reference for researchers and practitioners aiming to navigate the security aspects of QML.

Predominant Aspects on Security for Quantum Machine Learning: Literature Review

TL;DR

The paper addresses security in Quantum Machine Learning by performing a systematic literature review to map vulnerabilities and defenses in QML. It frames PQCs operating in a Hilbert space of dimension and analyzes how data encoding and hardware noise influence security, identifying non-classical attack vectors and upholding rigorous verification needs. Key contributions include a taxonomy of quantum-specific vulnerabilities (e.g., fault injections, cross-talk, Shuttles in ion traps) and defenses such as adversarial training, quantum differential privacy, MILP-based verification, and Lipschitz-based robustness measures, along with calls for cross-disciplinary benchmarking. The work offers guidance for researchers and practitioners seeking secure deployment of QML in real-world tasks and highlights the necessity of ongoing, experimentally validated security research in this rapidly evolving field.

Abstract

Quantum Machine Learning (QML) has emerged as a promising intersection of quantum computing and classical machine learning, anticipated to drive breakthroughs in computational tasks. This paper discusses the question which security concerns and strengths are connected to QML by means of a systematic literature review. We categorize and review the security of QML models, their vulnerabilities inherent to quantum architectures, and the mitigation strategies proposed. The survey reveals that while QML possesses unique strengths, it also introduces novel attack vectors not seen in classical systems. We point out specific risks, such as cross-talk in superconducting systems and forced repeated shuttle operations in ion-trap systems, which threaten QML's reliability. However, approaches like adversarial training, quantum noise exploitation, and quantum differential privacy have shown potential in enhancing QML robustness. Our review discuss the need for continued and rigorous research to ensure the secure deployment of QML in real-world applications. This work serves as a foundational reference for researchers and practitioners aiming to navigate the security aspects of QML.
Paper Structure (33 sections, 3 equations, 3 figures, 4 tables)

This paper contains 33 sections, 3 equations, 3 figures, 4 tables.

Table of Contents

  1. Introduction
  2. What is QML?
  3. Introduction to QC
  4. Brief history of computation
  5. Principles of QC
  6. State-of-the-art
  7. Introduction to QML
  8. Data encoding feature map
  9. Model zoo
  10. Hardware noise
  11. Research Methodology
  12. Filtering & Scoring
  13. Quantum Vulnerabilities
  14. Quantum attack vectors
  15. Fault Injections
  16. ...and 18 more sections

Figures (3)

  • Figure 1: Documentation of search process, adapted from prismaprisma
  • Figure 2: A taxonomy diagram detailing the key areas of vulnerabilities in QML.
  • Figure 3: A taxonomy diagram detailing the key areas and methodologies within the domain of defenses for QML.