CS405
Midterm Exam
The midterm exam covers everything up to and including Prolog. The test is open book, open notes, and open computer except for search engines and prolog. The test topics include:
Intro to AI
- Definition of AI
- What people are good at vs. what computers are good
at
- Major branches of AI
- Physical symbol hypothesis
- Turing Test
- Applications of AI
Non-Heuristic Search
- BFS, DFS, DFID (time, space, optimal, complete)
- Bidirectional search
- Dealing with duplicate states
Heuristic Search
- What a heuristic is, a heuristic vs. rules/algorithms
- Hill Climbing, simulated annealing, beam search
- Best First Search
- A* Search (use heuristic+actual cost)
- admissible heuristic, advantages, what it is
- Be prepared to demonstrate A* on a graph or other sort of
problem
- Minimax Search
- Common heuristics
- Alpha beta pruning
- Be prepared to identify states not examined in a
game tree with minimax and alpha-beta pruning
- Common issues
- Why a system might not make the direct move to
win
- Why ordering nodes to search may help improve pruning
- Machine Learning with Games (Samuel and Checkers)
- Rote learning
- Adjusting heuristic weights
Video Game AI
- Simplification in pathfinding
- Agents using finite state machines
- Creating a plan to achieve a goal
- Why video game AI uses simple techniques
- Subsumption architecture
Propositional/First Order Logic
- Map English sentences to a logic representation (mostly FOL) or vice versa
- Create some knowledge base in FOL
- Forward or Backward chaining on Horn Clauses
- No questions on Resolution
Prolog
- Represent facts in Prolog
- Write predicates that correspond to horn clauses (e.g. big(x) ^ green(x) -> hulk(x))
- Make recursive definitions; order matters, termination criteria first
- Make a query in prolog
- No questions requiring lists