Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Quantum information

Robust quantum classifiers via NISQ adversarial learning

Nature Computational Science volume 2, pages 699–700 (2022)Cite this article

Subjects

The vulnerability of quantum machine learning is demonstrated on a superconducting quantum computer, together with a defense strategy based on noisy intermediate-scale quantum (NISQ) adversarial learning.

Data-driven machine learning approaches are nowadays the most promising technique to make reliable predictions when the problem is too complex for mathematical modeling starting from first principles. Example applications include the classification of medical images and the development of self-driving cars. In these examples the data are purely classical, yet they can be loaded onto quantum registers to exploit the capabilities of quantum computers1. Alternatively, data can also be quantum, for example, when the task is to classify the states or the dynamical evolutions created in a quantum experiment. Example applications include quantum state and channel certification, classifying quantum phases of matter and quantum sensing2. The large state space allowed by quantum mechanics can provide quantum computers with computational advantage, for some very specific problems, but also poses some fundamental challenges: how can we efficiently extract information from this space? How sensitive is the result to small perturbations? Writing in this issue of Nature Computational Science, Wenhui Ren and colleagues3 have experimentally demonstrated in a 10-qubit superconducting quantum computer that quantum classifiers built using quantum neural networks and trained with conventional methods are vulnerable to adversarial attacks. Nonetheless, they have shown that this vulnerability can be significantly mitigated by using a different training technique called adversarial learning, which was experimentally tested with their device.

This is a preview of subscription content, access via your institution

Access options

Buy article

Get time limited or full article access on ReadCube.

$32.00

Buy

All prices are NET prices.

Fig. 1: Adversarial attack on a quantum classifier.

References

  1. Biamonte, J. et al. Nature 549, 195–202 (2017).

    Article  Google Scholar 

  2. Gebhart, V. et al. Preprint at https://arxiv.org/abs/2207.00298 (2022).

  3. Ren, W. et al., https://doi.org/10.1038/s43588-022-00351-9 (2022).

  4. Preskill, J. Quantum 2, 79 (2018).

    Article  Google Scholar 

  5. Liu, Y., Arunachalam, S. & Temme, K. Nat. Phys. 17, 1013–1017 (2021).

    Article  Google Scholar 

  6. Huang, H.-Y. et al. Science 376, 1182–1186 (2022).

    Article  MathSciNet  Google Scholar 

  7. Sharif, M., Bhagavatula, S., Bauer, L. & Reiter, M. K. In Proceedings of the 2016 ACM SIGSAC Conference on Computer and Communications Security 1528–1540 (Association for Computing Machinery, 2016).

  8. Wu, D., Xia, S.-T. & Wang, Y. Adv. Neural Inf. Process. Syst. 33, 2958–2969 (2020).

    Google Scholar 

  9. Banchi, L., Pereira, J. & Pirandola, S. PRX Quantum 2, 040321 (2021).

    Article  Google Scholar 

Download references

Author information

Authors and Affiliations

  1. Department of Physics and Astronomy, University of Florence, Florence, Italy

    Leonardo Banchi

  2. INFN Sezione di Firenze, Florence, Italy

    Leonardo Banchi

Authors
  1. Leonardo Banchi
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Leonardo Banchi.

Ethics declarations

Competing interests

The author declares no competing interests.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Banchi, L. Robust quantum classifiers via NISQ adversarial learning. Nat Comput Sci 2, 699–700 (2022). https://doi.org/10.1038/s43588-022-00359-1

Download citation

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/s43588-022-00359-1

Search

Advanced search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing