Bayesian Inference Reuse in Multilevel Models

We address efficient Bayesian inference in multilevel models, where group-specific latent variables are drawn from a shared hyperprior. In standard approaches, inferring the posterior for a new group requires revisiting all previous groups, incurring growing computational cost due to increased data volume and latent dimensionality. We propose replacing past groups with a set of weighted virtual observations of latent variables that preserve the prior over new groups, enabling fast, scalable inference. We provide theoretical analysis, empirical validation on case studies, and a reference implementation compatible with common probabilistic programming languages and inference algorithms.

• Publication year

  2026

• Venue

  ACM Transactions on Probabilistic Machine Learning

• Publication date

  2026-02-28

• Fields of study

  Not labeled

• Identifiers
  DOI 10.1145/3799415
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• The Bayesian Approach to Continual Learning: An Overview

  2025cited by this paper
• A survey on federated learning: challenges and applications

  2022cited by this paper
• A Critical Review of Communications in Multi-robot Systems

  2022cited by this paper
• Uncertain Evidence in Probabilistic Models and Stochastic Simulators

  2022cited by this paper
• Bayesian Coresets: Revisiting the Nonconvex Optimization Perspective

  2021cited by this paper
• Empirical Bayes

  2021cited by this paper
• Privacy Protection and Secondary Use of Health Data: Strategies and Methods

  2021cited by this paper
• A Survey on Stock Market Prediction Using Machine Learning Techniques

  2020cited by this paper
• Probabilistic Programs with Stochastic Conditioning

  2020cited by this paper
• Bayesian Pseudocoresets

  2020influential reference
• Gen: a general-purpose probabilistic programming system with programmable inference

  2019cited by this paper
• Sparse Variational Inference: Bayesian Coresets from Scratch

  2019influential reference
• Pyro: Deep Universal Probabilistic Programming

  2018cited by this paper
• Probabilistic programming with programmable inference

  2018cited by this paper
• Incremental inference for probabilistic programs

  2018cited by this paper
• Variational Continual Learning

  2017cited by this paper
• Stan: A Probabilistic Programming Language.

  2017cited by this paper
• Coresets for Scalable Bayesian Logistic Regression

  2016cited by this paper
• Inference Compilation and Universal Probabilistic Programming

  2016cited by this paper
• Design and Implementation of Probabilistic Programming Language Anglican

  2016cited by this paper
• Overcoming catastrophic forgetting in neural networks

  2016cited by this paper
• Deep Amortized Inference for Probabilistic Programs

  2016cited by this paper
• A New Approach to Probabilistic Programming Inference

  2014cited by this paper
• The logic of decision

  2014cited by this paper
• Just-In-Time Learning for Fast and Flexible Inference

  2014cited by this paper
• Bayesian data analysis, third edition

  2013influential reference
• The No-U-turn sampler: adaptively setting path lengths in Hamiltonian Monte Carlo

  2011cited by this paper
• Probabilistic reasoning in intelligent systems: Networks of plausible inference

  1991cited by this paper
• Socio-economic status and educational attainment: a multi-dimensional and multi-level study

  1991cited by this paper
• Bangladesh Fertility Survey 1989 (main report).

  1990cited by this paper
• Asymptotically Subminimax Solutions of Compound Statistical Decision Problems

  1985cited by this paper
• An Introduction to Empirical Bayes Data Analysis

  1985cited by this paper
• The Use of Historical Control Information in Testing for a Trend in Proportions

  1982cited by this paper
• Estimation in Parallel Randomized Experiments

  1981cited by this paper

• No citing papers are available for this paper.