Particle-based Variational Inference with Generalized Wasserstein Gradient Flow

Neural Information Processing Systems 

We show that GWG exhibits strong convergence guarantees.