Section: New Results

Axis 2: Pseudo-Bayesian Learning with Kernel Fourier Transform as Prior

Participants : Pascal Germain, Gael Letarte.

We revisit the kernel random Fourier features (RFF) method through the lens of the PAC-Bayesian theory. While the primary goal of RFF is to approximate a kernel, we look at the Fourier transform as a prior distribution over trigonometric hypotheses. It naturally suggests learning a posterior on these hypotheses. We derive generalization bounds that are optimized by learning a pseudo-posterior obtained from a closed-form expression, and corresponding learning algorithms This work has been accepted for publication at AISTATS 2019 conference [51].

It is a joint work with Emilie Morvant from Université Jean Monnet de Saint-Etienne.