Some sample exam 2 questions:

1. BRIEFLY define the following terms and give an example of how each term is used.
   • Valid (in logic)
   • Satisifiable (in logic)
   • Unsatisfiable (in logic)
   • Predicate
   • Function (in logic)
   • Propositional Logic
   • Model in First-order Logic
   • First-order Logic
   • Unification
   • Propositionalization
   • Conjuctive Normal Form
   • Horn Clause
   • Universal quantifier
   • Existential quantifier
   • Resolution
   • Forward Checking
   • Backward Checking
   • Situation Calculus
   • Frame Problem
   • Frame Axiom
   • Planning
   • Planning Action
   • STRIPS operators
   • Action Preconditions
   • Action Effects
   • Plan Goals
   • Interference
   • Threatening Actions
   • Ordering Actions
   • Inductive Learning
   • Decision Tree
   • Entropy
   • Information Gain
   • K-Nearest Neighbor
   • Curse of Dimensionality
   • Feature Normalization
   • Artificial Neural Network
   • Input Unit
   • Hidden Unit
   • Output Unit
   • Linear Perceptron
   • Thresholded Linear Perceptron
   • Sigmoid Function
   • Backpropagation
   • Linearly separable
2. Give algorithms for the following (in addition, for each algorithm indicate whether the search is complete, optimal and give some estimate of the cost of the algorithm):
   • Forward Chaining in propositional logic
   • Backward Chaining in propositional logic
   • Resolution in propositional logic
   • Forward Chaining in first-order logic
   • Backward Chaining in first-order logic
   • Resolution in first-order logic
   • Decision tree construction using information gain
   • K-nearest neighbor prediction
   • Learning in a simple linear perceptron
   • Backpropagation
3. For each of the algorithms discussed above define a set of propositions/predicates and then show how the algorithm would work for those rules/facts.
4. Define the rule Modus Ponens for propositional logic. Give an example. How does it differ for first-order logic?
5. Repeat the previous question for Resolution.
6. How would you convert a logical formula to Conjunctive Normal Form.
7. How would you convert a logical formula to Horn Clauses.
8. Explain how first-order logic differs from propositional logic. Give an example of the key parts of first-order logic.
9. Why is First-Order Logic used over Propositional Logic?
10. Look at samples of logical sentences in the text and convert them to first-order logic.
11. How do the universal, existential quantifiers and negation relate?
12. What is the frame problem? Why is it important in planning? What is a frame axiom? Give an example.
13. Give two notions of negation in logic. Explain the advantages and disadvantages of each.
14. What is unification? Why is it so important to first-order logic?
15. How does planning differ from search? Give an example.
16. How do STRIPS operators work? How are they used in planning?
17. How does interference occur in planning? Give an example. What are some ways to avoid it?
18. Decision trees are often chosen as a learning method because they are easy to understand and are complete in the set of hypotheses they can learn. What is meant by the latter?
19. Artificial neural networks are said to be "inspired" by human brains? In what sense? Why do we focus on the human brain?
20. What is the importance of the sigmoid function in backpropagation learning?
21. Many learning methods such as the k-nearest neighbor method rely on normalization to be effective. Explain why this is and give an example demonstrating why normalization would be needed.