Practical Probabilistic Programming

mind-and-cognition

• A model has four ingredients: variables, dependencies, functional forms, numerical parameters — a complete grammar for representing uncertain belief.

  PPA-RC-081

• The set of variables assigned values in a program can differ from run to run — probabilistic programs exceed Bayesian networks in expressive power.

  PPA-RC-094

• By observing sunny days in a month, you infer whether someone is likely to have a good mood — reasoning flows from evidence to hidden state.

  PPA-RC-046

education-and-formation

• When you make numeric arguments elements themselves, you get compound versions — uncertainty about uncertainty, modeled naturally.

  PPA-RC-040

• You'll step back for a deeper understanding of probabilistic models and inference, then explore creative ways to build rich, practical models.

  PPA-RC-071

work-as-meaning

• Building a probabilistic model requires four main ingredients: elements, dependencies, functional forms, and parameters — engineering as epistemology.

  PPA-RC-060