CS 289, Fall 2004, Stuart Russell
Weekly Questions for Knowledge Representation and Reasoning
Weeks 1 and 2 (due 9/8): AI, knowledge-based systems, propositional logic
- In the commentary on McCarthy's "Programs with Common Sense", what is Bar-Hillel's
main criticism of McCarthy's proposal? In retrospect, was it valid?
- What major element is missing from McCarthy's deductive formulation of planning?
- Do Exercises 7.4(c) and (e) from AIMA2e.
- Briefly explain the key elements that make CHAFF efficient.
- Do you think Schöning's algorithm is likely to be more or less efficient than WalkSAT in practice?
Week 3 (due 9/13): QBFs, propositional agents.
- Explain very roughly how a chess problem such as "Can White mate in two?"
corresponds to a QBF.
(You need not say exactly what all the propositional variables are.)
What about Kriegspiel, where the black pieces are invisible?
- Do Exercise 7.13(a) and (b) from AIMA2e.
- According to Rosenschein and Kaelbling, what
is the role of knowledge representation and reasoning in intelligent agents?
Could an R-K agent use explicitly represented knowledge, and, if so, how?
Week 4 (due 9/20): First-order logic: representation
- Do AIMA2e Ex. 8.9, and include a formal proof of the answer to the question.
- Do AIMA2e Ex. 9.1.
- Explain why first-order logic might be preferable to propositional logic for knowledge representation.
Week 6 (due 10/4): First-order logic: inference
- Do AIMA2e Ex. 9.8.
- Do AIMA2e Ex. 9.20.
- Explain why PTTP is (a) complete and (b) fairly efficient.
Week 7 (due 10/11): Reasoning about action
- Do AIMA2e Ex. 10.3.
- Do AIMA2e Ex. 11.15.
- Do AIMA2e Ex. 11.17.
Week 8 (due 10/18): Reasoning about action contd.; general ontology.
- Do AIMA2e Ex. 12.17.
- What do Bertoli et al. use OBDDS for and why?
- Do AIMA2e Ex. 10.10.
Week 9 (due 10/25): General ontologies contd.; reasoning about knowledge and communication.
No questions for this week.
Week 10 (due 11/5): Bayesian networks: representation and inference.
- Do AIMA2e Ex. 14.8.
- Do AIMA2e Ex. 14.9.
- (optional) It may be helpful to get some extra practice by doing
Exs. 14.5 and 14.10 also.
Week 11 (due 11/8): Bayesian networks contd.
No questions for this week.
Week 12 (due 11/15): Qualitative probabilistic networks; temporal probability models.
- Explain how a QPN decision model could be used for reasoning
by analogy, i.e., appealing to a past decided case to obtain a decision on
a new case.
- Do AIMA2e Ex. 15.2.
- Do AIMA2e Ex. 15.3.
- Do AIMA2e Ex. 15.9.
Week 13 (due 11/22): Temporal probability models contd: activity monitoring and robotics.
Week 14 (due 11/29): First-order probabilistic languages
Week 15 (due 12/6): First-order probabilistic languages contd.