Differentiable and Probabilistic Programming
Probabilistic programming is a method for writing statistical models by writing programs.
The idea of differentiable programming is to consider programs for which the derivative can be found automatically. This has applications to machine learning and statistical inference.