You will do well on the final exam if you can do the following:
1. Write a set of STRIPS operators for a given problem domain. Hint: suppose Shakey can push objects around?
2. Given a set of STRIPS operators and a particular state in a search space, calculate the value of the heuristic obtained by relaxing the problem, using the automatically derived heuristics described in the lecture on heuristics.
3. Calculate by hand the alpha and beta values in a small game tree, and determine the part of the game tree that would be pruned by alpha-beta pruning.
4. Derive the learning rule for a single sigmoid neuron by calculating the derivative of the sum of squared error function.
5. Do all the problems on the logical exercises handout. Be sure you understand how the problem of encoding N-Queens using clause schemas works.
6. Do all the problems on the probabilistic exercises handout. Be sure you understand how to determine if two variables are conditionally independent in a Bayesian network.
7. By hand, perform the steps Gibbs sampling would perform on a small Bayesian network. Hint: how can Gibbs sampling get "stuck" and always get the same value for a variable when it is sampled?
8. Do all the problems on the midterm exam.