BriefGPT.xyz
Jun, 2019
近似推断将深度网络转化为高斯过程
Approximate Inference Turns Deep Networks into Gaussian Processes
HTML
PDF
Mohammad Emtiyaz Khan, Alexander Immer, Ehsan Abedi, Maciej Korzepa
TL;DR
本文研究神经网络和高斯过程之间的关系，证明了贝叶斯神经网络的高斯后验近似等同于高斯过程的后验。在训练神经网络时，利用高斯过程的边缘似然函数来调整神经网络的超参数，得到的核函数是神经切向核。我们的工作旨在促进进一步将神经网络和高斯过程在实际应用中相结合的研究。
Abstract
deep neural networks
(DNN) and
gaussian processes
(GP) are two powerful models with several theoretical connections relating them, but the relationship between their training methods is not well understood. In th
→