CS475/CS575: What you need to know for the Final exam.
Chapter 2
- Know the distinction between a performance measure and
a utility function.
- Know the definition and some of the strengths and weaknesses of the four types of agents
discussed: simple reflex, model-based agents,
goal-based agents and utility-based agents.
Chapter 3
- Know some definitions: initial state, successor function,
state space, search tree, path, goal test,
path cost, step cost, optimal solution.
- What are the criteria for evaluating search strategies? completeness,
time complexity, space complexity, and optimality.
- Know the characteristics of breadth-first search, uniform cost search,
depth-first search, depth-limited search,
iterative deepening, and bidirectional search.
Be able to show order in which nodes are expanded and contents of the queue
after each step of search algorithms.
Chapter 4
- Know what greedy best-first search is.
- Know what A* search is, its optimality, completeness.
Be able to show order in which nodes are expanded and contents of the queue
after each step for greedy and A*.
Know what makes a heuristic function h admissible.
What is the correlation between domination of heuristics and efficiency of A*?
- What is hill climbing? What are some of its problems?
Know the algorithm.
- What is simulated annealing?
Chapter 5
- Know what a constraint satisfaction problem (CSP) is.
Be able to represent problems as CSPs (variables, domains, constraints).
Be able to draw constraint graphs for binary CSPs.
- What are some of the search strategies
used? backtracking search, forward checking, arc consistency, local search with
min-conflicts heuristic. Be able to solve CSPs using these strategies.
- What are some heuristics for constraint satisfaction problems?
minimum remaining values / degree heuristic, least-constraining value.
- If a constraint graph forms a tree, how can a solution be found in time linear
in the number of variables? Be able to use the algorithm.
Chapter 6
- Understand minimax.
- What are some problems that develop when you have to cut off search
before you reach an end state?
- Understand alpha-beta pruning. Given an arbitrary game tree,
be able to show what
branches would be pruned by alpha-beta cutoffs.
- Understand how chance affects the minimax algorithm.
Chapter 7
- What are the syntax and semantics of a knowledge representation
language? What is entailment, model, logical inference?
When is an inference procedure sound? complete?
- Know the syntax and semantics of propositional logic.
- Know standard logical equivalences.
- Definitions: valid, satisfiable, unsatisfiable.
- Be able to show a sentence is valid, satisfiable or unsatisfiable
using a truth table or equivalences.
Be able to use Modus Ponens and Resolution inference rules.
- What is conjunctive normal form (CNF)?
Be able to convert a sentence into a CNF.
- Be able to use forward and backward chaining algorithms to
determine determine whether a query is entailed by a knowledge base.
Chapter 8
- Know the syntax and semantics of first-order logic.
Know definitions of model, interpretation, term, atomic and complex sentence,
connections between quantifiers, equality.
- Be able to write a first-order logic sentence that corresponds
to a given English sentence.
- Here are some samples:
- "John is a farmer."
- "Everybody loves Frank Sinatra."
- "Everybody whose age is over 50 loves a crooner."
Chapter 9
- Understand unification. Be able to compute a most general unifier.
- Be able to use Generalized Modus Ponens.
- Understand how forward and backward chaining work.
- Using resolution and proof by refutation: Be able to take a KB of
first-order logic sentences, convert them to CNF form, and use resolution
to show that a particular sentence is entailed by the knowledge base.
- Be able to show that a sentence is valid or unsatisfiable using resolution refutation procedure.
Chapter 11
- Understand partial-order planning. Know how to graphically
illustrate a plan (causal and temporal links). When a plan is a solution?
when it is consistent and has no open preconditions.
- Be able to draw a partial order plan for a problem,
give linearizations of a partial-order plan.