Bayesian Networks and Markov

Question 1

What is a Graphical model, and what are the two types?

Answer 1

A natural way of modeling random variables and predictions.
Two types:
- Undirected graph (markov network): can be cyclic or not
- Directed graph (bayesian):
- inference is performed according to bayes rule
- acyclic by nature

Question 2

What is a Markovian blanket?

Answer 2

A markovian blanket is when the parents, children and parents of children of a node are known. This makes the node independent. q

Question 3

Explain Bayesian filtering.

Answer 3

You approximate the distribution of random variables in a Bayesian network. Can be used for object tracking in a Bayesian network.

Question 4

Explain Hidden Markov model.

Answer 4

The simplest form of Dynamic Bayesian Network. You have two nodes at each instance, one hidden and one known.

Question 5

What are some ways that machine learning can be used in computer vision?

Answer 5

• Classification
• Segmentation
• Object detection
• Pose estimation