Skip to main content
Springer Nature Link
Log in

Detecting Model Misspecification in Amortized Bayesian Inference with Neural Networks

  • Conference paper
  • First Online:
Pattern Recognition (DAGM GCPR 2023)

Part of the book series: Lecture Notes in Computer Science ((LNCS,volume 14264))

Included in the following conference series:

  • 724 Accesses

Abstract

Recent advances in probabilistic deep learning enable efficient amortized Bayesian inference in settings where the likelihood function is only implicitly defined by a simulation program (simulation-based inference; SBI). But how faithful is such inference if the simulation represents reality somewhat inaccurately—that is, if the true system behavior at test time deviates from the one seen during training? We conceptualize the types of model misspecification arising in SBI and systematically investigate how the performance of neural posterior approximators gradually deteriorates under these misspecifications, making inference results less and less trustworthy. To notify users about this problem, we propose a new misspecification measure that can be trained in an unsupervised fashion (i.e., without training data from the true distribution) and reliably detects model misspecification at test time. Our experiments clearly demonstrate the utility of our new measure both on toy examples with an analytical ground-truth and on representative scientific tasks in cell biology, cognitive decision making, and disease outbreak dynamics. We show how the proposed misspecification test warns users about suspicious outputs, raises an alarm when predictions are not trustworthy, and guides model designers in their search for better simulators.

This is a preview of subscription content, log in via an institution to check access.

Access this chapter

Log in via an institution

Subscribe and save

Springer+ Basic
$34.99 /Month
  • Get 10 units per month
  • Download Article/Chapter or eBook
  • 1 Unit = 1 Article or 1 Chapter
  • Cancel anytime
Subscribe now

Buy Now

Chapter
USD 29.95
Price excludes VAT (USA)
  • Available as PDF
  • Read on any device
  • Instant download
  • Own it forever

Tax calculation will be finalised at checkout

Purchases are for personal use only

Institutional subscriptions

Similar content being viewed by others

Notes

  1. 1.

    We demonstrate in Experiment 1 that model misspecification also affects the performance of non-amortized sequential neural posterior estimation.

References

  1. Alquier, P., Ridgway, J.: Concentration of tempered posteriors and of their variational approximations. arXiv:1706.09293 [cs, math, stat] (2019). arXiv: 1706.09293

  2. Ardizzone, L., Lüth, C., Kruse, J., Rother, C., Köthe, U.: Guided image generation with conditional invertible neural networks (2019)

    Google Scholar 

  3. Berger, J.O., Wolpert, R.L.: The Likelihood Principle. No. v. 6 in Lecture Notes-Monograph Series. 2nd edn. Institute of Mathematical Statistics, Hayward (1988)

    Google Scholar 

  4. Bieringer, S., et al.: Measuring QCD splittings with invertible networks. SciPost Phys. Proc. 10(6), 126 (2021)

    Google Scholar 

  5. Bloem-Reddy, B., Teh, Y.W.: Probabilistic symmetries and invariant neural networks. J. Mach. Learn. Res. 21, 90–1 (2020)

    MathSciNet  Google Scholar 

  6. Butter, A., et al.: Machine learning and LHC event generation. arXiv preprint arXiv:2203.07460 (2022)

  7. Bürkner, P.C., Gabry, J., Vehtari, A.: Approximate leave-future-out cross-validation for Bayesian time series models. J. Stat. Comput. Simul. 90(14), 2499–2523 (2020). https://doi.org/10.1080/00949655.2020.1783262. arXiv:1902.06281 [stat]

  8. Cannon, P., Ward, D., Schmon, S.M.: Investigating the impact of model misspecification in neural simulation-based inference (2022). arXiv:2209.01845 [cs, stat]

  9. Cranmer, K., Brehmer, J., Louppe, G.: The frontier of simulation-based inference. Proc. Natl. Acad. Sci. 117(48), 30055–30062 (2020)

    Article  MathSciNet  Google Scholar 

  10. Dehning, J., et al.: Inferring change points in the spread of COVID-19 reveals the effectiveness of interventions. Science 369(6500) (2020)

    Google Scholar 

  11. Delaunoy, A., Hermans, J., Rozet, F., Wehenkel, A., Louppe, G.: Towards reliable simulation-based inference with balanced neural ratio estimation (2022). arXiv:2208.13624 [cs, stat]

  12. Dellaporta, C., Knoblauch, J., Damoulas, T., Briol, F.X.: Robust Bayesian inference for simulator-based models via the MMD Posterior Bootstrap (2022). https://doi.org/10.48550/ARXIV.2202.04744

  13. Dong, E., Du, H., Gardner, L.: An interactive web-based dashboard to track COVID-19 in real time. Lancet. Infect. Dis 20(5), 533–534 (2020). https://doi.org/10.1016/S1473-3099(20)30120-1

    Article  Google Scholar 

  14. Durkan, C., Murray, I., Papamakarios, G.: On contrastive learning for likelihood-free inference. In: International Conference on Machine Learning, pp. 2771–2781. PMLR (2020)

    Google Scholar 

  15. Frazier, D.T., Drovandi, C.: Robust approximate Bayesian inference with synthetic likelihood. J. Comput. Graph. Stat. 30(4), 958–976 (2021). https://doi.org/10.1080/10618600.2021.1875839

    Article  MathSciNet  Google Scholar 

  16. Frazier, D.T., Robert, C.P., Rousseau, J.: Model misspecification in approximate Bayesian computation: consequences and diagnostics. J. Royal Stat. Soc. Ser. B (Stat. Method.) 82(2), 421–444 (2020). https://doi.org/10.1111/rssb.12356

    Article  MathSciNet  Google Scholar 

  17. Gabry, J., Simpson, D., Vehtari, A., Betancourt, M., Gelman, A.: Visualization in Bayesian workflow. J. Royal Stat. Soc. Ser. A (Stat. Soc.) 182(2), 389–402 (2019)

    Google Scholar 

  18. Ghaderi-Kangavari, A., Rad, J.A., Nunez, M.D.: A general integrative neurocognitive modeling framework to jointly describe EEG and decision-making on single trials. Comput. Brain Behav. (2023). https://doi.org/10.1007/s42113-023-00167-4

    Article  Google Scholar 

  19. Giummolè, F., Mameli, V., Ruli, E., Ventura, L.: Objective Bayesian inference with proper scoring rules. TEST 28(3), 728–755 (2019)

    Article  MathSciNet  Google Scholar 

  20. Gonçalves, P.J., et al.: Training deep neural density estimators to identify mechanistic models of neural dynamics. Elife 9, e56261 (2020)

    Article  Google Scholar 

  21. Greenberg, D., Nonnenmacher, M., Macke, J.: Automatic posterior transformation for likelihood-free inference. In: International Conference on Machine Learning, pp. 2404–2414. PMLR (2019)

    Google Scholar 

  22. Gretton, A., Borgwardt, K., Rasch, M., Schölkopf, B., Smola, A.: A Kernel two-sample test. J. Mach. Learn. Res. 13, 723–773 (2012)

    Google Scholar 

  23. Grünwald, P., Van Ommen, T., et al.: Inconsistency of Bayesian inference for misspecified linear models, and a proposal for repairing it. Bayesian Anal. 12(4), 1069–1103 (2017)

    Article  MathSciNet  Google Scholar 

  24. Hermans, J., Begy, V., Louppe, G.: Likelihood-free MCMC with amortized approximate ratio estimators. In: International Conference on Machine Learning, pp. 4239–4248. PMLR (2020)

    Google Scholar 

  25. Hermans, J., Delaunoy, A., Rozet, F., Wehenkel, A., Louppe, G.: Averting a crisis in simulation-based inference. arXiv preprint arXiv:2110.06581 (2021)

  26. Holmes, C.C., Walker, S.G.: Assigning a value to a power likelihood in a general Bayesian model. Biometrika 104(2), 497–503 (2017)

    MathSciNet  Google Scholar 

  27. Jones-Todd, C.M., et al.: Identifying prognostic structural features in tissue sections of colon cancer patients using point pattern analysis: Point pattern analysis of colon cancer tissue sections. Stat. Med. 38(8), 1421–1441 (2019). https://doi.org/10.1002/sim.8046

    Article  MathSciNet  Google Scholar 

  28. Knoblauch, J., Jewson, J., Damoulas, T.: Generalized variational inference: three arguments for deriving new posteriors. arXiv preprint arXiv:1904.02063 (2019)

  29. von Krause, M., Radev, S.T., Voss, A.: Mental speed is high until age 60 as revealed by analysis of over a million participants. Nat. Hum. Behav. 6(5), 700–708 (2022). https://doi.org/10.1038/s41562-021-01282-7

    Article  Google Scholar 

  30. Leclercq, F.: Simulation-based inference of Bayesian hierarchical models while checking for model misspecification (2022). arXiv:2209.11057 [astro-ph, q-bio, stat]

  31. Loaiza-Maya, R., Martin, G.M., Frazier, D.T.: Focused Bayesian prediction. J. Appl. Economet. 36(5), 517–543 (2021)

    Article  MathSciNet  Google Scholar 

  32. Lotfi, S., Izmailov, P., Benton, G., Goldblum, M., Wilson, A.G.: Bayesian model selection, the marginal likelihood, and generalization. arXiv preprint arXiv:2202.11678 (2022)

  33. Lueckmann, J.M., Boelts, J., Greenberg, D., Goncalves, P., Macke, J.: Benchmarking simulation-based inference. In: International Conference on Artificial Intelligence and Statistics, pp. 343–351. PMLR (2021)

    Google Scholar 

  34. Lueckmann, J.M., Boelts, J., Greenberg, D., Goncalves, P., Macke, J.: Benchmarking simulation-based inference. In: Banerjee, A., Fukumizu, K. (eds.) Proceedings of The 24th International Conference on Artificial Intelligence and Statistics. Proceedings of Machine Learning Research, vol. 130, pp. 343–351. PMLR (2021)

    Google Scholar 

  35. Lueckmann, J.M., Goncalves, P.J., Bassetto, G., Öcal, K., Nonnenmacher, M., Macke, J.H.: Flexible statistical inference for mechanistic models of neural dynamics. In: Advances in Neural Information Processing Systems, vol. 30 (2017)

    Google Scholar 

  36. Masegosa, A.: Learning under model misspecification: applications to variational and ensemble methods. In: Advances in Neural Information Processing Systems, vol. 33, pp. 5479–5491 (2020)

    Google Scholar 

  37. Matsubara, T., Knoblauch, J., Briol, F.X., Oates, C.J.: Robust generalised bayesian inference for intractable likelihoods (2022). arXiv:2104.07359 [math, stat]

  38. Pacchiardi, L., Dutta, R.: Likelihood-free inference with generative neural networks via scoring rule minimization. arXiv preprint arXiv:2205.15784 (2022)

  39. Pacchiardi, L., Dutta, R.: Score matched neural exponential families for likelihood-free inference (2022). arXiv:2012.10903 [stat]

  40. Pang, G., Shen, C., Cao, L., Hengel, A.V.D.: Deep learning for anomaly detection: a review. ACM Comput. Surv. 54(2), 1–38 (2022). https://doi.org/10.1145/3439950. arXiv:2007.02500 [cs, stat]

  41. Papamakarios, G., Murray, I.: Fast \(\varepsilon \)-free inference of simulation models with Bayesian conditional density estimation. In: Advances in Neural Information Processing Systems, vol. 29 (2016)

    Google Scholar 

  42. Papamakarios, G., Pavlakou, T., Murray, I.: Masked autoregressive flow for density estimation. In: Advances in Neural Information Processing Systems, vol. 30 (2017)

    Google Scholar 

  43. Papamakarios, G., Sterratt, D., Murray, I.: Sequential neural likelihood: fast likelihood-free inference with autoregressive flows. In: The 22nd International Conference on Artificial Intelligence and Statistics, pp. 837–848. PMLR (2019)

    Google Scholar 

  44. Radev, S.T., et al.: OutbreakFlow: model-based Bayesian inference of disease outbreak dynamics with invertible neural networks and its application to the COVID-19 pandemics in Germany. PLoS Comput. Biol. 17(10), e1009472 (2021)

    Article  Google Scholar 

  45. Radev, S.T., Mertens, U.K., Voss, A., Ardizzone, L., Köthe, U.: BayesFlow: learning complex stochastic models with invertible neural networks. IEEE Trans. Neural Netw. Learn. Syst. 33, 1452–1466 (2020)

    Google Scholar 

  46. Ramesh, P., et al.: GATSBI: generative adversarial training for simulation-based inference. arXiv preprint arXiv:2203.06481 (2022)

  47. Schmon, S.M., Cannon, P.W., Knoblauch, J.: Generalized posteriors in approximate Bayesian computation (2021). arXiv:2011.08644 [stat]

  48. Shiono, T.: Estimation of agent-based models using Bayesian deep learning approach of BayesFlow. J. Econ. Dyn. Control 125, 104082 (2021)

    Article  MathSciNet  Google Scholar 

  49. Säilynoja, T., Bürkner, P.C., Vehtari, A.: Graphical test for discrete uniformity and its applications in goodness of fit evaluation and multiple sample comparison (2021). arXiv:2103.10522 [stat]

  50. Talts, S., Betancourt, M., Simpson, D., Vehtari, A., Gelman, A.: Validating Bayesian inference algorithms with simulation-based calibration (2020). arXiv:1804.06788 [stat]

  51. Tejero-Cantero, A., et al.: SBI-a toolkit for simulation-based inference. arXiv preprint arXiv:2007.09114 (2020)

  52. Thomas, O., Corander, J.: Diagnosing model misspecification and performing generalized Bayes’ updates via probabilistic classifiers. arXiv preprint arXiv:1912.05810 (2019)

  53. Vehtari, A., Ojanen, J.: A survey of Bayesian predictive methods for model assessment, selection and comparison. Stat. Surv. 6 (2012). https://doi.org/10.1214/12-SS102

  54. Ward, D., Cannon, P., Beaumont, M., Fasiolo, M., Schmon, S.M.: Robust neural posterior estimation and statistical model criticism (2022). arXiv:2210.06564 [cs, stat]

  55. White, H.: Maximum likelihood estimation of misspecified models. Econometrica 50(1), 1–25 (1982)

    Article  MathSciNet  Google Scholar 

  56. Wiqvist, S., Frellsen, J., Picchini, U.: Sequential neural posterior and likelihood approximation. arXiv preprint arXiv:2102.06522 (2021)

  57. Yang, J., Zhou, K., Li, Y., Liu, Z.: Generalized out-of-distribution detection: a survey. arXiv:2110.11334 (2021)

  58. Zhang, F., Gao, C.: Convergence rates of variational posterior distributions. Ann. Stat. 48(4), 2180–2207 (2020). https://doi.org/10.1214/19-AOS1883

Download references

Acknowledgments

MS and PCB were supported by the Cyber Valley Research Fund (grant number: CyVy-RF-2021-16) and the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany’s Excellence Strategy EXC-2075 - 390740016 (the Stuttgart Cluster of Excellence SimTech). UK was supported by the Informatics for Life initiative funded by the Klaus Tschira Foundation. STR was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany’s Excellence Strategy - EXC-2181 - 390900948 (the Heidelberg Cluster of Excellence STRUCTURES).

Author information

Authors and Affiliations

  1. Cluster of Excellence SimTech, University of Stuttgart, Stuttgart, Germany

    Marvin Schmitt & Paul-Christian Bürkner

  2. Department of Statistics, TU Dortmund University, Dortmund, Germany

    Paul-Christian Bürkner

  3. Visual Learning Lab, Heidelberg University, Heidelberg, Germany

    Ullrich Köthe

  4. Cluster of Excellence STRUCTURES, Heidelberg University, Heidelberg, Germany

    Stefan T. Radev

Authors
  1. Marvin Schmitt
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Paul-Christian Bürkner
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Ullrich Köthe
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Stefan T. Radev
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Marvin Schmitt .

Editor information

Editors and Affiliations

  1. IWR, Heidelberg University, Heidelberg, Germany

    Ullrich Köthe

  2. IWR, Heidelberg University, Heidelberg, Germany

    Carsten Rother

1 Electronic supplementary material

Below is the link to the electronic supplementary material.

Supplementary material 1 (pdf 3496 KB)

Rights and permissions

Reprints and permissions

Copyright information

© 2024 The Author(s), under exclusive license to Springer Nature Switzerland AG

About this paper

Check for updates. Verify currency and authenticity via CrossMark

Cite this paper

Schmitt, M., Bürkner, PC., Köthe, U., Radev, S.T. (2024). Detecting Model Misspecification in Amortized Bayesian Inference with Neural Networks. In: Köthe, U., Rother, C. (eds) Pattern Recognition. DAGM GCPR 2023. Lecture Notes in Computer Science, vol 14264. Springer, Cham. https://doi.org/10.1007/978-3-031-54605-1_35

Download citation

  • DOI: https://doi.org/10.1007/978-3-031-54605-1_35

  • Published:

  • Publisher Name: Springer, Cham

  • Print ISBN: 978-3-031-54604-4

  • Online ISBN: 978-3-031-54605-1

  • eBook Packages: Computer ScienceComputer Science (R0)

Publish with us

Policies and ethics