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.

Lessons on interpretable machine learning from particle physics

Nature Reviews Physics volume 4, pages 284–286 (2022)Cite this article

Subjects

Machine learning methods have proved powerful in particle physics, but without interpretability there is no guarantee the outcome of a learning algorithm is correct or robust. Christophe Grojean, Ayan Paul, Zhuoni Qian and Inga Strümke give an overview of how to introduce interpretability to methods commonly used in particle physics.

For more than a decade now, machine learning (ML) algorithms have played a central role in discoveries at the Large Hadron Collider, such as the discovery of the Higgs boson and matter–antimatter asymmetry1. Although the outcomes of these analyses have been well understood, the inner workings of the ML models have remained obscure because of the inexplicability of the parameter space or topology of the models. This black-box approach to statistical analyses has limited the rate at which ML has been applied to a wider range of physics problems, because interpretability is of prime importance. Under a black-box approach, the predictive power of a ML model does not guarantee that the mathematical structure statistically modelled by the learning algorithm corresponds to the dynamics driving a physical process. Of the many possible configurations of the parametric solutions of the ML algorithm, several may be unphysical or simply be correct in only a subset of the states that the actual physical system can explore. Hence, the knowledge of how and why a ML model predicts a certain outcome for a physical system provides confidence that the results are not, by fluke, only seemingly correct, but rather stand on robust grounds of well-understood physical theories. Such interpretability is key even when the underlying physics is known (that is, in the standard model), but especially when the laws of new physics beyond the standard model have to be inferred from data collected at experiments.

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

Relevant articles

Open Access articles citing this article.

Access options

Rent or buy this article

Get just this article for as long as you need it

$39.95

Learn more

Prices may be subject to local taxes which are calculated during checkout

Fig. 1: 12-variable machine learning-assisted analysis for classifying five particle-production channels.

References

  1. Guest, D., Cranmer, K. & Whiteson, D. Deep Learning and its Application to LHC Physics. Ann. Rev. Nucl. Part. Sci. 68, 161–181 (2018).

    Article  ADS  Google Scholar 

  2. Murdoch, W. J., Singh, C., Kumbier, K., Abbasi-Asl, R. & Yu, B. Definitions, methods, and applications in interpretable machine learning. PNAS 116, 22071–22080 (2019).

    Article  MathSciNet  Google Scholar 

  3. Barredo Arrieta, A. et al. Explainable artificial intelligence (xai): Concepts, taxonomies, opportunities and challenges toward responsible AI. Inf. Fusion 58, 82–115 (2020).

    Article  Google Scholar 

  4. Hamon, R., Junklewitz, H. & Sanchez, I. Robustness and explainability of Artificial Intelligence. Publ. Off. Eur. Union, Luxembourg (2020).

  5. Rudin, C. Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead. Nat. Mach. Intell. 1, 206–215 (2019).

    Article  Google Scholar 

  6. Chen, T. & Guestrin, C. XGBoost: A scalable tree boosting system. In Proc. 22nd ACM SIGKDD Int. Conf. Knowledge Discovery and Data Mining (KDD ‘16) 785–794 (ACM, 2016).

  7. Fawagreh, K., Gaber, M. M. & Elyan, E. Random forests: from early developments to recent advancements. Syst. Sci. Control. Eng. 2, 602–609 (2014).

    Article  Google Scholar 

  8. Ribeiro, M. T., Singh, S. & Guestrin, S. “Why should I trust you?”: Explaining the predictions of any classifier. In Proc. 22nd ACM SIGKDD Int. Conf. Knowledge Discovery and Data Mining (KDD ‘16) 1135–1144 (ACM, 2016).

  9. Lundberg, S. M. et al. From local explanations to global understanding with explainable AI for trees. Nat. Mach. Intell. 2, 56–67 (2020).

    Article  Google Scholar 

  10. Grojean, C., Paul, A. & Qian, Z. Resurrecting \(b\bar{b}h\) with kinematic shapes. J. High Energy Phys. 4, 139 (2021).

    Article  ADS  Google Scholar 

Download references

Acknowledgements

This work benefited from support by the Deutsche Forschungsgemeinschaft under Germany’s Excellence Strategy EXC 2121 “Quantum Universe”–390833306. The work of A.P. is funded by Volkswagen Foundation within the initiative “Corona Crisis and Beyond–Perspectives for Science, Scholarship and Society”. I.S. is grateful to the Norwegian Research Council for support through the EXAIGON project–Explainable AI systems for gradual industry adoption (grant no. 304843).

Author information

Authors and Affiliations

  1. Deutsches Elektronen-Synchrotron DESY, Hamburg, Germany

    Christophe Grojean & Ayan Paul

  2. Institut für Physik, Humboldt Universität zu Berlin, Berlin, Germany

    Christophe Grojean & Ayan Paul

  3. School of Physics, Hangzhou Normal University, Hangzhou, Zhejiang, China

    Zhuoni Qian

  4. Department of Computer Science, Norwegian University of Science and Technology, Trondheim, Norway

    Inga Strümke

Authors
  1. Christophe Grojean
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Ayan Paul
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Zhuoni Qian
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Inga Strümke
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Ayan Paul.

Ethics declarations

Competing interests

The authors declare no competing interests.

Rights and permissions

Reprints and Permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Grojean, C., Paul, A., Qian, Z. et al. Lessons on interpretable machine learning from particle physics. Nat Rev Phys 4, 284–286 (2022). https://doi.org/10.1038/s42254-022-00456-0

Download citation

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/s42254-022-00456-0

This article is cited by

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