BriefGPT.xyz
May, 2024
任意非线性贝叶斯神经网络的小样本变分推断
Few-sample Variational Inference of Bayesian Neural Networks with Arbitrary Nonlinearities
HTML
PDF
David J. Schodt
TL;DR
本研究通过仅使用3个确定性样本来传播统计矩,实现了对任意非线性网络层的统计矩传播,从而使得少样本变分推断成为可能,并将此方法应用于一种新的非线性激活函数,用于向Bayesian神经网络的输出节点注入物理先验信息。
Abstract
bayesian neural networks
(BNNs) extend traditional neural networks to provide
uncertainties
associated with their outputs. On the forward pass through a BNN, predictions (and their
→