Chapter 5. Modeling dependencies with Bayesian and Markov networks
This chapter covers
- Types of relationships among variables in a probabilistic model and how these relationships translate into dependencies
- How to express these various types of dependencies in Figaro
- Bayesian networks: models that encode directed dependencies among variables
- Markov networks: models that encode undirected dependencies among variables
- Practical examples of Bayesian and Markov networks
In chapter 4, you learned about the relationships between probabilistic models and probabilistic programs, and you also saw the ingredients of a probabilistic model, which are variables, dependencies, functional forms, and numerical parameters. This chapter focuses on two modeling frameworks: Bayesian networks and Markov networks. Each framework is based on a different way of encoding dependencies.