# AIMA - Probabilistic Reasoning

In which we explain how to build efficient network models to reason under uncertainty according to the laws of probability theory, and how to distinguish between correlation and causality.

## Flashcards

What is a Bayesian network?

An efficient representation of full joint probability distributions that contain information about the relationships between random variables.

Informally, what is a DAG?

A directed tree but nodes are allowed to have multiple parents.

In a Bayesian network, what does $X \to Y$ represent?

$X$ has a direct effect on $Y$.

What is a hybrid Bayesian network?

A Bayes net containing both discrete and continuous variables.

Why is exact inference in Bayesian networks exponential in the worst case?

Because you can encode SAT problems into Bayesian networks.

What two things does a query to a Bayesian network contain ?

- A set of query variables, probabilities you’re asking for.
- A set of evidence variables.

How can you make approximate inference in Bayesian networks 2-3 OOM faster?

Compile the Bayesian network into a small program.

What is a causal network?

Bayesian networks where only causal relationships are allowed.

What is the benefit of using a causal network rather than a Bayesian network?

You can predict the effects of interventions.