用大模型生成语义上下文点，构建更优的不确定性估计框架
Function-Space Empirical Bayes Regularisation with Large Vision-Language Model Priors
- 用大视觉语言模型生成语义上下文点，构建函数空间先验
- 在数据稀疏和分布外场景下提升预测性能与不确定性可靠性
- 适合关注模型置信度与鲁棒性的研究者使用
贝叶斯深度学习（BDL）通过结合深度神经网络与贝叶斯推断，为可靠不确定性量化提供了原则性框架。其核心挑战在于设计能有效扩展至高维数据的信息性先验分布。近期的函数变分推断（VI）方法通过在函数空间直接施加先验来应对这一问题；然而，多数现有方法依赖高斯过程（GP）先验，在高维情形下表达能力和泛化能力受限。本文提出VLM-FS-EB，一种新型函数空间经验贝叶斯正则化框架，利用大视觉语言模型（VLMs）生成语义有意义的上下文点，并将其用于构建表达性强的函数先验。实验表明，该方法在多种基线对比中持续提升预测性能，尤其在分布外（OOD）检测任务和数据稀缺场景下表现出更可靠的不确定性估计。
原文摘要 · Abstract (English)
Bayesian deep learning (BDL) provides a principled framework for reliable uncertainty quantification by combining deep neural networks with Bayesian inference. A central challenge in BDL lies in the design of informative prior distributions that scale effectively to high-dimensional data. Recent functional variational inference (VI) approaches address this issue by imposing priors directly in function space; however, most existing methods rely on Gaussian process (GP) priors, whose expressiveness and generalisation capabilities become limited in high-dimensional regimes. In this work, we propose VLM-FS-EB, a novel function-space empirical Bayes regularisation framework, leveraging large vision-language models (VLMs) to generates semantically meaningful context points. These synthetic samples are then used VLMs for embeddings to construct expressive functional priors. Furthermore, the proposed method is evaluated against various baselines, and experimental results demonstrate that our method consistently improves predictive performance and yields more reliable uncertainty estimates, particularly in out-of-distribution (OOD) detection tasks and data-scarce regimes.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。