arXivDaily arXiv每日学术速递 周一至周五更新
北美退伍军人节(Veterans Day),12 日无更新
arXiv 2006.01490stat.MLastro-ph.IMcs.LG
收藏

贝叶斯神经网络

Bayesian Neural Networks

CNRS(法国国家科学研究中心);Institut d'Astrophysique de Paris(巴黎天体物理研究所);Mila - Quebec Artificial Intelligence Institute(米拉-魁北克人工智能研究所);Flatiron Institute(熨斗研究所);CEA(法国原子能和替代能源委员会);World Scientific Publishing Co(世界科学出版公司)

Tom Charnock, Laurence Perreault-Levasseur, François Lanusse

摘要:

近年来,神经网络已成为分析复杂和抽象数据模型的强大工具。然而,它们的引入本质上增加了我们的不确定性,即哪些分析特征是与模型相关的,哪些是由于神经网络。这意味着神经网络的预测具有偏差,这些偏差不能简单地与数据创建和观察的真实性质区分开来。为了试图解决这些问题,我们讨论贝叶斯神经网络:神经网络的不确定性,由于网络可以被表征。特别是,我们提出了贝叶斯统计框架,使我们能够根据观察某些数据的固有随机性和我们缺乏关于如何创建和观察数据的知识的不确定性来对不确定性进行分类。在提出这样的技术,我们展示了如何预测误差的神经网络可以获得的原则,并提供了两个有利的方法来表征这些错误。我们还将描述这两种方法在实践中如何存在实质性的缺陷,强调在使用神经网络时需要其他统计技术才能真正进行推理。

英文摘要:

In recent times, neural networks have become a powerful tool for the analysis of complex and abstract data models. However, their introduction intrinsically increases our uncertainty about which features of the analysis are model-related and which are due to the neural network. This means that predictions by neural networks have biases which cannot be trivially distinguished from being due to the true nature of the creation and observation of data or not. In order to attempt to address such issues we discuss Bayesian neural networks: neural networks where the uncertainty due to the network can be characterised. In particular, we present the Bayesian statistical framework which allows us to categorise uncertainty in terms of the ingrained randomness of observing certain data and the uncertainty from our lack of knowledge about how data can be created and observed. In presenting such techniques we show how errors in prediction by neural networks can be obtained in principle, and provide the two favoured methods for characterising these errors. We will also describe how both of these methods have substantial pitfalls when put into practice, highlighting the need for other statistical techniques to truly be able to do inference when using neural networks.

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。

↑