Bayesian Mixture of Experts For Large Language Models

• URL: http://arxiv.org/abs/2511.08968v1
• Date: Thu, 13 Nov 2025 01:22:24 GMT
• Title: Bayesian Mixture of Experts For Large Language Models
• Authors: Maryam Dialameh, Hossein Rajabzadeh, Weiwei Zhang, Walid Ahmed, Hyock Ju Kwon,
• Abstract summary: We present a post-hoc uncertainty estimation framework for large language models (LLMs) based on Mixture-of-Experts architectures.<n>Bayesian-MoE applies a structured Laplace approximation to the second linear layer of each expert, enabling calibrated uncertainty estimation.<n> Experiments on common-sense reasoning benchmarks with Qwen1.5-MoE and DeepSeek-MoE demonstrate that Bayesian-MoE improves both expected calibration error (ECE) and negative log-likelihood (NLL) over baselines.
• Score: 2.889541910837398
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: We present Bayesian Mixture of Experts (Bayesian-MoE), a post-hoc uncertainty estimation framework for fine-tuned large language models (LLMs) based on Mixture-of-Experts architectures. Our method applies a structured Laplace approximation to the second linear layer of each expert, enabling calibrated uncertainty estimation without modifying the original training procedure or introducing new parameters. Unlike prior approaches, which apply Bayesian inference to added adapter modules, Bayesian-MoE directly targets the expert pathways already present in MoE models, leveraging their modular design for tractable block-wise posterior estimation. We use Kronecker-factored low-rank approximations to model curvature and derive scalable estimates of predictive uncertainty and marginal likelihood. Experiments on common-sense reasoning benchmarks with Qwen1.5-MoE and DeepSeek-MoE demonstrate that Bayesian-MoE improves both expected calibration error (ECE) and negative log-likelihood (NLL) over baselines, confirming its effectiveness for reliable downstream decision-making.

Related papers

• Improving Minimax Estimation Rates for Contaminated Mixture of Multinomial Logistic Experts via Expert Heterogeneity [49.809923981964715]
  Contaminated mixture of experts (MoE) is motivated by transfer learning methods where a pre-trained model, acting as a frozen expert, is integrated with an adapter model, functioning as a trainable expert, in order to learn a new task.<n>In this work, we characterize uniform convergence rates for estimating parameters under challenging settings where ground-truth parameters vary with the sample size.<n>We also establish corresponding minimax lower bounds to ensure that these rates are minimax optimal.
  arXiv  Detail & Related papers  (2026-01-31T23:45:50Z)
• 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)
• Extracting Uncertainty Estimates from Mixtures of Experts for Semantic Segmentation [9.817102014355617]
  We show that well-calibrated predictive uncertainty estimates can be extracted from a mixture of experts (MoE) without architectural modifications.<n>Our results show that MoEs yield more reliable uncertainty estimates than ensembles in terms of conditional correctness metrics.<n>Our experiments on the Cityscapes dataset suggest that increasing the number of experts can further enhance uncertainty calibration.
  arXiv  Detail & Related papers  (2025-09-05T05:30:53Z)
• On Equivariant Model Selection through the Lens of Uncertainty [49.137341292207]
  Equivariant models leverage prior knowledge on symmetries to improve predictive performance, but misspecified architectural constraints can harm it instead.<n>We compare frequentist (via Conformal Prediction), Bayesian (via the marginal likelihood), and calibration-based measures to naive error-based evaluation.<n>We find that uncertainty metrics generally align with predictive performance, but Bayesian model evidence does so inconsistently.
  arXiv  Detail & Related papers  (2025-06-23T13:35:06Z)
• Embedded Nonlocal Operator Regression (ENOR): Quantifying model error in learning nonlocal operators [8.585650361148558]
  We propose a new framework to learn a nonlocal homogenized surrogate model and its structural model error. This framework provides discrepancy-adaptive uncertainty quantification for homogenized material response predictions in long-term simulations.
  arXiv  Detail & Related papers  (2024-10-27T04:17:27Z)
• Self-Evolutionary Large Language Models through Uncertainty-Enhanced Preference Optimization [9.618391485742968]
  Iterative preference optimization has recently become one of the de-facto training paradigms for large language models (LLMs) We present an uncertainty-enhanced textbfPreference textbfOptimization framework to make the LLM self-evolve with reliable feedback. Our framework substantially alleviates the noisy problem and improves the performance of iterative preference optimization.
  arXiv  Detail & Related papers  (2024-09-17T14:05:58Z)
• Beyond mirkwood: Enhancing SED Modeling with Conformal Predictions [0.0]
  We propose an advanced machine learning-based approach that enhances flexibility and uncertainty in SED fitting. We incorporate conformalized quantile regression to convert point predictions into error bars, enhancing interpretability and reliability.
  arXiv  Detail & Related papers  (2023-12-21T11:27:20Z)
• Local Bayesian Dirichlet mixing of imperfect models [0.0]
  We study the ability of Bayesian model averaging and mixing techniques to mine nuclear masses. We show that the global and local mixtures of models reach excellent performance on both prediction accuracy and uncertainty quantification.
  arXiv  Detail & Related papers  (2023-11-02T21:02:40Z)
• Calibrating Neural Simulation-Based Inference with Differentiable Coverage Probability [50.44439018155837]
  We propose to include a calibration term directly into the training objective of the neural model. By introducing a relaxation of the classical formulation of calibration error we enable end-to-end backpropagation. It is directly applicable to existing computational pipelines allowing reliable black-box posterior inference.
  arXiv  Detail & Related papers  (2023-10-20T10:20:45Z)
• Measuring and Modeling Uncertainty Degree for Monocular Depth Estimation [50.920911532133154]
  The intrinsic ill-posedness and ordinal-sensitive nature of monocular depth estimation (MDE) models pose major challenges to the estimation of uncertainty degree. We propose to model the uncertainty of MDE models from the perspective of the inherent probability distributions. By simply introducing additional training regularization terms, our model, with surprisingly simple formations and without requiring extra modules or multiple inferences, can provide uncertainty estimations with state-of-the-art reliability.
  arXiv  Detail & Related papers  (2023-07-19T12:11:15Z)
• Learning to Estimate Without Bias [57.82628598276623]
  Gauss theorem states that the weighted least squares estimator is a linear minimum variance unbiased estimation (MVUE) in linear models. In this paper, we take a first step towards extending this result to non linear settings via deep learning with bias constraints. A second motivation to BCE is in applications where multiple estimates of the same unknown are averaged for improved performance.
  arXiv  Detail & Related papers  (2021-10-24T10:23:51Z)
• Providing reliability in Recommender Systems through Bernoulli Matrix Factorization [63.732639864601914]
  This paper proposes Bernoulli Matrix Factorization (BeMF) to provide both prediction values and reliability values. BeMF acts on model-based collaborative filtering rather than on memory-based filtering. The more reliable a prediction is, the less liable it is to be wrong.
  arXiv  Detail & Related papers  (2020-06-05T14:24:27Z)
• Efficient Ensemble Model Generation for Uncertainty Estimation with Bayesian Approximation in Segmentation [74.06904875527556]
  We propose a generic and efficient segmentation framework to construct ensemble segmentation models. In the proposed method, ensemble models can be efficiently generated by using the layer selection method. We also devise a new pixel-wise uncertainty loss, which improves the predictive performance.
  arXiv  Detail & Related papers  (2020-05-21T16:08:38Z)

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.