BriefGPT.xyz
Oct, 2023
用哈密尔顿蒙特卡洛估计最优的PAC-Bayes界限
Estimating optimal PAC-Bayes bounds with Hamiltonian Monte Carlo
HTML
PDF
Szilvia Ujváry, Gergely Flamich, Vincent Fortuin, José Miguel Hernández Lobato
TL;DR
该论文通过对MNIST数据集进行实验，研究了PAC-Bayes参数约束为分解高斯分布时在优化PAC-Bayes界限时可能损失的紧密度，结果表明在某些情况下存在5-6%的显著紧密度差距。
Abstract
An important yet underexplored question in the
pac-bayes
literature is how much tightness we lose by restricting the posterior family to
factorized gaussian distributions
when optimizing a
→