SGD for Variational Inference: Tackling Unbounded Variance via Preconditioning and Dynamic Batching

• URL: http://arxiv.org/abs/2605.07531v1
• Date: Fri, 08 May 2026 10:02:51 GMT
• Title: SGD for Variational Inference: Tackling Unbounded Variance via Preconditioning and Dynamic Batching
• Abstract summary: Black-Box Variational Inference (BBVI) typically relies on Gradient Descent (SGD) to optimize Lower Bound (ELBO)<n>We bridge the gap between foundational optimization theory and the practical instances of BBVI.
• Score: 8.6165142481624
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Black-Box Variational Inference (BBVI) typically relies on Stochastic Gradient Descent (SGD) to optimize the Evidence Lower Bound (ELBO). However, the stochastic gradients in BBVI inherently exhibit unbounded variance, violating standard assumptions and instead satisfying the weaker Blum-Gladyshev (BG) condition, where variance grows quadratically with distance from the optimum. In this paper, we bridge the gap between stochastic optimization theory and the practical instances of BBVI. Focusing on the broad elliptic location-scale family of parameterized distributions, we offer two main contributions. First, we prove the existence of an ELBO solution, a foundational property usually assumed a priori in the literature. Second, we establish comprehensive convergence guarantees spanning finite-time and asymptotic regimes for Minibatch Projected SGD (PSGD) equipped with dynamic batching and preconditioning under the BG condition. Our theoretical framework demonstrates that dynamic batching combined with preconditioning systematically enables rigorous guarantees even in complex settings. We illustrate our theoretical findings with numerical results, highlighting the efficacy of our approach for modern inference tasks.

Related papers

• Too Correct to Learn: Reinforcement Learning on Saturated Reasoning Data [55.84428098924793]
  We propose Constrained Uniform Top-K Sampling (CUTS), a parameter-free decoding strategy enforcing structure-preserving exploration.<n>We integrate this into Mixed-CUTS, a training framework synergizing exploitative and exploratory rollouts to amplify intra-group advantage variance.<n> Notably, Mixed-CUTS improves Pass@1 accuracy on the challenging AIME25 benchmark by up to 15.1% over standard GRPO.
  arXiv  Detail & Related papers  (2026-04-20T16:43:28Z)
• The Procrustean Bed of Time Series: The Optimization Bias of Point-wise Loss [53.542743390809356]
  This paper aims to provide a first-principles analysis of the Expectation of Optimization Bias (EOB)<n>Our analysis reveals a fundamental paradigm paradox: the more deterministic and structured the time series, the more severe the bias by point-wise loss function.<n>We present a concrete solution that simultaneously achieves both principles via DFT or DWT.
  arXiv  Detail & Related papers  (2025-12-21T06:08:22Z)
• Bridging the Gap Between Bayesian Deep Learning and Ensemble Weather Forecasts [100.26854618129039]
  Weather forecasting is fundamentally challenged by the chaotic nature of the atmosphere.<n>Recent advances in Bayesian Deep Learning (BDL) offer a promising but often disconnected alternative.<n>We bridge these paradigms through a unified hybrid BDL framework for ensemble weather forecasting.
  arXiv  Detail & Related papers  (2025-11-18T07:49:52Z)
• Federated Generalised Variational Inference: A Robust Probabilistic Federated Learning Framework [12.454538785810259]
  FedGVI is a probabilistic Federated Learning (FL) framework that is robust to both prior and likelihood misspecification.<n>We offer theoretical analysis in terms of fixed-point convergence, optimality of the cavity distribution, and provable robustness to likelihood misspecification.
  arXiv  Detail & Related papers  (2025-02-02T16:39:37Z)
• Batch and match: black-box variational inference with a score-based divergence [26.873037094654826]
  We propose batch and match (BaM) as an alternative approach to blackbox variational inference (BBVI) based on a score-based divergence. We show that BaM converges in fewer evaluations than leading implementations of BBVI based on ELBO.
  arXiv  Detail & Related papers  (2024-02-22T18:20:22Z)
• Generalized Schrödinger Bridge Matching [54.171931505066]
  Generalized Schr"odinger Bridge (GSB) problem setup is prevalent in many scientific areas both within and without machine learning. We propose Generalized Schr"odinger Bridge Matching (GSBM), a new matching algorithm inspired by recent advances. We show that such a generalization can be cast as solving conditional optimal control, for which variational approximations can be used.
  arXiv  Detail & Related papers  (2023-10-03T17:42:11Z)
• On the Convergence of Black-Box Variational Inference [16.895490556279647]
  We provide the first convergence guarantee for full black-box variational inference (BBVI) Our results hold for log-smooth posterior densities with and without strong log-concavity and the location-scale variational family.
  arXiv  Detail & Related papers  (2023-05-24T16:59:50Z)
• Practical and Matching Gradient Variance Bounds for Black-Box Variational Bayesian Inference [8.934639058735812]
  We show that BBVI satisfies a matching bound corresponding to the $ABC$ condition used in the gradient descent literature. We also show that the variance of the mean-field parameterization has provably superior dimensional dependence.
  arXiv  Detail & Related papers  (2023-03-18T19:07:14Z)
• Towards Extremely Fast Bilevel Optimization with Self-governed Convergence Guarantees [42.514612465664605]
  We propose a single-level formulation to uniformly understand existing explicit and implicit Gradient-based BLOs. A striking feature of our convergence result is that, compared to those original unaccelerated GBLO versions, the fast BAGDC admits a unified non-asymptotic convergence theory towards stationarity.
  arXiv  Detail & Related papers  (2022-05-20T09:46:10Z)
• BCD Nets: Scalable Variational Approaches for Bayesian Causal Discovery [97.79015388276483]
  A structural equation model (SEM) is an effective framework to reason over causal relationships represented via a directed acyclic graph (DAG) Recent advances enabled effective maximum-likelihood point estimation of DAGs from observational data. We propose BCD Nets, a variational framework for estimating a distribution over DAGs characterizing a linear-Gaussian SEM.
  arXiv  Detail & Related papers  (2021-12-06T03:35:21Z)
• Regularizing Variational Autoencoder with Diversity and Uncertainty Awareness [61.827054365139645]
  Variational Autoencoder (VAE) approximates the posterior of latent variables based on amortized variational inference. We propose an alternative model, DU-VAE, for learning a more Diverse and less Uncertain latent space.
  arXiv  Detail & Related papers  (2021-10-24T07:58:13Z)
• Posterior Differential Regularization with f-divergence for Improving Model Robustness [95.05725916287376]
  We focus on methods that regularize the model posterior difference between clean and noisy inputs. We generalize the posterior differential regularization to the family of $f$-divergences. Our experiments show that regularizing the posterior differential with $f$-divergence can result in well-improved model robustness.
  arXiv  Detail & Related papers  (2020-10-23T19:58:01Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.