### Final exam topics

You should be prepared to answer the following sorts of questions:
- Definition questions - for example, "What is bias in a learning
algorithm?"
- "Trace the algorithm" questions. For example, "Here are some
sample documents - show the Naive Bayes calculations."
- Synthesis questions that ask you to tie together different topics
from the class, or describe when a particular algorithm, assumption,
or approach is valid.

In addition to the topics from midterm
1 and midterm 2, you should be
familiar with:
- Probability
- Bayes' rule and its application
- Naive Bayes
- Inference with the joint probability distribution
- Construction of Bayesian networks
- Inference in Bayesian Networks

- Utility
- Expected utility calculations
- Value of information calculations
- Policies
- Value Iteration
- Policy Iteration

- Learning
- K-means clustering
- N-fold cross-validation
- Q-learning (and its relationship to value iteration)
- Perceptrons
- Multilayer networks and backpropagation