arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2603.27488cs.LG
收藏

变分学习分数后验

Variational Learning of Fractional Posteriors

DSO National Laboratories, Singapore(新加坡DSO国家实验室)

Kian Ming A. Chai, Edwin V. Bonilla

AI总结:

本文提出一种新的单参数变分目标,用于下界数据证据并估计近似分数后验,通过扩展框架实现分层构造和贝叶斯后验,展示了在混合模型中的模拟研究,证明分数后验能获得更好的校准效果。

摘要:

我们介绍了一种新的单参数变分目标,该目标下界数据证据并使近似分数后验的估计成为可能。我们扩展了这一框架,用于分层构造和贝叶斯后验,提供了一种灵活的概率建模工具。我们展示了两种情况,其中梯度可以解析地获得,并在混合模型的模拟研究中证明,我们的分数后验可以比传统变分界估计的后验获得更好的校准。当应用于变分自编码器(VAEs)时,我们的方法能够获得更高的证据界,并使高表现的近似贝叶斯后验与分数后验联合学习。我们证明了使用分数后验训练的VAEs生成器在从先验生成时具有更好的对齐性。

英文摘要:

We introduce a novel one-parameter variational objective that lower bounds the data evidence and enables the estimation of approximate fractional posteriors. We extend this framework to hierarchical construction and Bayes posteriors, offering a versatile tool for probabilistic modelling. We demonstrate two cases where gradients can be obtained analytically and a simulation study on mixture models showing that our fractional posteriors can be used to achieve better calibration compared to posteriors from the conventional variational bound. When applied to variational autoencoders (VAEs), our approach attains higher evidence bounds and enables learning of high-performing approximate Bayes posteriors jointly with fractional posteriors. We show that VAEs trained with fractional posteriors produce decoders that are better aligned for generation from the prior.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑