All times as US Eastern Time.
Despite remarkable progress over the last two decades in reducing causal inference to statistical practice, the "causal revolution" proclaimed by Judea Pearl and other pioneers remains incomplete, with a sprawling and fragmented technical literature that is still inaccessible to non-experts and isolated from the cutting-edge computational methods and software tools being developed within mainstream machine learning research. Probabilistic programming languages are promising substrates for bridging this gap thanks to the close correspondence between their operational semantics and most standard mathematical formalisms for causal inference, especially that of structural causal models. This talk will introduce ChiRho (https://github.com/BasisResearch/chirho), a new causal probabilistic programming language embedded in Python. ChiRho extends an existing probabilistic programming language (Pyro) that is built on algebraic effects and handlers with new algebraic operations for performing interventions and counterfactuals on causal models represented as probabilistic programs, and new effect handlers for automatically reducing causal inference computations over these models to ordinary probabilistic computations on transformed probabilistic programs. I will also illustrate ChiRho's design with representative example applications from econometrics, single-cell biology, and epidemiology.
Universal probabilistic programming systems allow users to construct complex models whose support varies between different executions. Unfortunately, this poses a substantial challenge for their underlying inference engines, with most common inference schemes not applicable to this difficult setting. In this talk, we will discuss two inference schemes for effectively dealing with such problems by decomposing the original program into a mixture over sub-programs with static support. We will then show that such programs implicitly define a Bayesian model average over paths, and reveal why this can lead to pathologies when performing full Bayesian inference. To address this, we will introduce some alternative schemes for weighting the different program paths that provide more robust prediction under model misspecification.
Non-parametric statistics is incredibly expressive (Gaussian processes, Dirichlet processes, and beyond). Several authors have proposed probabilistic programming as a convenient language for describing non-parametric models—more convenient than traditional mathematical notation. In this talk I will discuss this, and some recent progress, both theoretical and practical.