Grassmann Stein Variational Gradient Descent
This work addresses a bottleneck in particle inference algorithms for high-dimensional Bayesian inference, offering an incremental improvement over existing SVGD variants.
The paper tackles the problem of variance underestimation in Stein variational gradient descent (SVGD) for high-dimensional target distributions by proposing Grassmann Stein variational gradient descent (GSVGD), which allows projections onto arbitrary dimensional subspaces and updates projectors simultaneously, resulting in efficient state-space exploration in high-dimensional problems with low-dimensional structure.
Stein variational gradient descent (SVGD) is a deterministic particle inference algorithm that provides an efficient alternative to Markov chain Monte Carlo. However, SVGD has been found to suffer from variance underestimation when the dimensionality of the target distribution is high. Recent developments have advocated projecting both the score function and the data onto real lines to sidestep this issue, although this can severely overestimate the epistemic (model) uncertainty. In this work, we propose Grassmann Stein variational gradient descent (GSVGD) as an alternative approach, which permits projections onto arbitrary dimensional subspaces. Compared with other variants of SVGD that rely on dimensionality reduction, GSVGD updates the projectors simultaneously for the score function and the data, and the optimal projectors are determined through a coupled Grassmann-valued diffusion process which explores favourable subspaces. Both our theoretical and experimental results suggest that GSVGD enjoys efficient state-space exploration in high-dimensional problems that have an intrinsic low-dimensional structure.