1: \begin{abstract}

2:   Particle-based variational inference methods (ParVIs) have gained attention in the Bayesian inference literature, for their capacity to yield flexible and accurate approximations.

3:   We explore ParVIs from the perspective of Wasserstein gradient flows, and make both theoretical and practical contributions.

4:   We unify various finite-particle approximations that existing ParVIs use, and recognize that the approximation is essentially a compulsory smoothing treatment, in either of two equivalent forms.

5:   This novel understanding reveals the assumptions and relations of existing ParVIs, and also inspires new ParVIs. % with two novel instances.

6:   We propose an acceleration framework and a principled bandwidth-selection method for general ParVIs; these are based on the developed theory and leverage the geometry of the Wasserstein space.

7:   Experimental results show the improved convergence by the acceleration framework and enhanced sample accuracy by the bandwidth-selection method.

8: \end{abstract}

9: