Inference as risk identification
Open Access
- 11 September 2026
- journal article
- research article
- Published by Springer Nature in Computational Management Science
- Vol. 23 (2) , 34
Abstract
The Bayes posterior solves the variational problem of maximizing the expected likelihood of the data minus the Kullback–Leibler divergence from the prior distribution. We generalize Bayesian inference from this variational point of view. In particular, we introduce a variational inference (VI) problem whose solutions we call posteriors and show that these posteriors are risk identifiers, i.e., subgradients of risk measures. This insight leads to: ( i ) rigorous results concerning the existence and asymptotics of posteriors, ( ii ) the ability to constrain the divergence from the prior—which motivates a measure-valued relaxation of M-estimation that we call coherent inference —and ( iii ) insights into VI problems derived from $$\phi$$ -divergences. These insights include semi-analytical expressions for posteriors and connections with utility theory. We use the semi-analytical expressions for $$\phi$$ -divergence posteriors to numerically solve a multi-armed bandit problem.
Keywords
Funding Information
- Sandia National Laboratories LDRD “Risk-Adaptive Experimental Design for High-Consequence Systems”
- U.S. Department of Energy Office of Science Early Career Research Program
- Norges Forskningsråd (SURE-AI: The Norwegian Centre for Sustainable, Risk-Averse, and Ethical AI grant 357482)
This publication has 52 references indexed in Scilit:
- Foundations of Symmetric Spaces of Measurable Functions2016
- The Safe Bayesian2012
- The Elements of Statistical Learning2009
- A Course in Functional Analysis2007
- Perturbation Analysis of Optimization Problems2000
- Variational Analysis1998
- A Decision-Theoretic Generalization of On-Line Learning and an Application to BoostingJournal of Computer and System Sciences, 1997
- Minimum disparity estimation for continuous models: Efficiency, distributions and robustnessAnnals of the Institute of Statistical Mathematics, 1994
- On the convergence of subdifferentials of convex functionsArchiv der Mathematik, 1993
- Asymptotic evaluation of certain markov process expectations for large time, ICommunications on Pure and Applied Mathematics, 1975