Estimating optimal PAC-Bayes bounds with Hamiltonian Monte Carlo

Szilvia Ujváry, Gergely Flamich, Vincent Fortuin, José Miguel Hernández-Lobato

Keywords: PAC-Bayes bounds, MCMC, Hamiltonian Monte Carlo, generalization, deep learning

TL;DR: We estimate data-independent PAC-Bayes bounds using Hamiltonian Monte Carlo samples from the optimal (Gibbs) posterior and achieve better risk certificates than MFVI.

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 PAC-Bayes bound. We investigate this issue by estimating data-independent PAC-Bayes bounds using the optimal posteriors, comparing them to bounds obtained using MFVI. Concretely, we (1) sample from the optimal Gibbs posterior using Hamiltonian Monte Carlo, (2) estimate its KL divergence from the prior with thermodynamic integration, and (3) propose three methods to obtain high-probability bounds under different assumptions. Our experiments on the MNIST dataset reveal significant tightness gaps, as much as 5-6% in some cases.

Submission Number: 11