Annealed Stein Variational Gradient Descent
This addresses a key limitation in approximate inference methods for researchers and practitioners, though it is an incremental improvement over SVGD.
The paper tackled the problem of Stein variational gradient descent (SVGD) struggling with multi-modal distributions, specifically particles getting stuck in local modes and failing to reproduce density across regions, and proposed an annealing schedule that significantly improved mode coverage in experiments.
Particle based optimization algorithms have recently been developed as sampling methods that iteratively update a set of particles to approximate a target distribution. In particular Stein variational gradient descent has gained attention in the approximate inference literature for its flexibility and accuracy. We empirically explore the ability of this method to sample from multi-modal distributions and focus on two important issues: (i) the inability of the particles to escape from local modes and (ii) the inefficacy in reproducing the density of the different regions. We propose an annealing schedule to solve these issues and show, through various experiments, how this simple solution leads to significant improvements in mode coverage, without invalidating any theoretical properties of the original algorithm.