In collaboration with Gideon Dresdner, Samuel Harrison and Yu Yao, I implemented the paper Boosting Black Box Variational Inference using Pyro. We have contributed a tutorial to the Pyro documentation to illustrate how to use our implementation.
To quote from the tutorial: “This tutorial demonstrates how to implement boosting black box Variational Inference in Pyro. In boosting Variational Inference, we approximate a target distribution with an iteratively selected mixture of densities. In cases where a single density provided by regular Variational Inference doesn’t adequately approximate a target density, boosting VI thus offers a simple way of getting more complex approximations. We show how this can be implemented as a relatively straightforward extension of Pyro’s SVI.”