Introduction to Automatic Backward Filtering Forward Guiding
share
In this document I aim to give an informal treatment of automatic Backward Filtering Forward Guiding, a general algorithm for conditional sampling from a Markov process on a directed acyclic graph. I'll show that the underlying ideas can be understood with a basic background in probability and statistics. The more technical treatment is the paper Automatic backward filtering forward guiding for Markov processes and graphical models (Van der Meulen and Schauer, 2021). I specifically assume some background knowledge on likelihood based inference and Bayesian statistics. The final sections are more demanding and assume familiarity with continuous-time stochastic processes constructed from their infinitesimal generator. Clearly, all work discussed here is the result of research carried out over the past decade together with various collaborators, most importantly Moritz Schauer (Chalmers University of Technology and University of Gothenburg, Sweden). Section 8 is based on joint work with Marcin Mider (Trium Analysis Online GmbH, Germany) and Frank Schäfer (University of Basel, Switzerland) as well.
READ FULL TEXT
