Study Sheet for Quiz 2

=

Artificial Intelligence: Problem Solving (presentation on AI: Search)

How can problems or puzzles be captured as search problems? What does a search tree look like? What is the difference between breadth-first search and depth-first search? What is the difference between uninformed (blind) search and informed (heuristic) search? What is a pretty simple heuristic to use in search? You should know how to create the search tree for each of those methods for such puzzles as the cannibals and missionaries problem, the 8-puzzle, and the water jugs problem. Can all problems be captured as search problems? What kinds of problems are AI's good at, and what kinds of problems are they not good at? What may be some differences between how humans solve problems and how problems are solved using search trees?

Artificial Intelligence: Reasoning (presentation on AI: Logic)

How can formal logic be used in AI? What is Automated Theorem Proving? What are some limitations of the current state of the art in Automated Theorem Proving? What is Prolog, and how does it work?

Artificial Intelligence: Decision Trees (see Decision Trees Website. Also look at this guide)

What is a decision tree? What explains the general slogan: the smaller the tree, the better? Informally, how does the 'gain' splitting function work, and what is its purpose? What is the importance of having a testing set in addition to a training set?

Artificial Intelligence: Learning (presentation on AI: Neural Networks)

What are the basic components of a connectionist network? What is the general dynamics of a connectionist network? Specifically, how does the Jets and Sharks network work, what can you do with it, and what explains some of its behaviors? What is the word-advantage effect, and how can the Word-recognition neural network explain this effect? How was the face recognition network trained to achieve good performance? What are some intereresting features of connectionist networks pertaining to cognition?