mlsanityData to 7 Oct 2026

PACm-Bayes: Narrowing the Empirical Risk Gap in the Misspecified Bayesian Regime

arXiv 2010.09629 ↗First posted 19 Oct 2020cs.LG stat.MLcited by 15

Not on the map. Why?

Abstract (arXiv)

The Bayesian posterior minimizes the "inferential risk" which itself bounds the "predictive risk". This bound is tight when the likelihood and prior are well-specified. However since misspecification induces a gap, the Bayesian posterior predictive distribution may have poor generalization performance. This work develops a multi-sample loss (PACm) which can close the gap by spanning a trade-off between the two risks. The loss is computationally favorable and offers PAC generalization guarantees. Empirical study demonstrates improvement to the predictive distribution.

Key facts. Cited by 15 papers in the map (data through 7 Oct 2026). First posted on arXiv 19 Oct 2020.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 3; 2022: 5; 2023: 3; 2024: 2; 2025: 0; 2026: 2.

15 papers.

Sort:

  1. 2021Priors in Bayesian Deep Learning: A Reviewcited by 101
  2. 2022Bayesian Model Selection, the Marginal Likelihood, and Generalizationcited by 56
  3. 2021Dangers of Bayesian Model Averaging under Covariate Shiftcited by 31
  4. 2023Quantum Conformal Prediction for Reliable Uncertainty Quantification in Quantum Machine Learningcited by 11
  5. 2022Robust Bayesian Learning for Reliable Wireless AI: Framework and Applicationscited by 9
  6. 2022Robust PACm: Training Ensemble Models Under Misspecification and Outlierscited by 6
  7. 2024Calibrating Bayesian Learning via Regularization, Confidence Minimization, and Selective Inferencecited by 6
  8. 2022Cold Posteriors through PAC-Bayescited by 3
  9. 2022On the Performance of Direct Loss Minimization for Bayesian Neural Networkscited by 3
  10. 2023The fine print on tempered posteriorscited by 3
  11. 2024Predictive variational inference: Learn the predictively optimal posterior distributioncited by 3
  12. 2021Uncertainty estimation under model misspecification in neural network regressioncited by 2
3 more
  1. 2023If there is no underfitting, there is no Cold Posterior Effectcited by 2
  2. 2026Variational Learning of Fractional Posteriorscited by 0
  3. 2026Beliefs Beyond Posteriors: The Predictive Cost of ELBO Bindingcited by 0