CPSC 453, Fall 2003 Information About Test 2 ------------------------------------------------------------------------- The second test for this course will be given in class on Monday, March 24. It covers all the course work that we have done since the previous test. This includes expert systems and a bit of planning (Chapter 7 of the text) and MiniMax (pages 144 to 155). We also read two stories by Jorge Luis Borges and an excerpt from "Understanding Computers and Cognition." You should be aware of the issues raised by these readings. In addition, you should have some basic familiarity with the CLIPS expert system shell. Although we will begin Chapter 9 (machine learning) this week, machine learing is NOT covered on this test. Here are some of the important topics for the test: production systems condition/action rules working memory recognize/act cycle conflict set expert systems knowledge base inference engine explanation subsystem knowledge engineering brittleness rule-based expert systems (production systems) explanation in rule-based expert systems data-driven operation of rule-based systems ("forward chaining") can be depth-first, breadth-first, best-first goal-driven operation of rule-based systems Prolog as a kind of goal-driven production system advantages and disadvantages of rule-based expert systems CLIPS facts and rules asserting and retracting facts using defrule to define a rule patterns and pattern-matching variables; anonymous variables the agenda, rule activation, and rule firing search strategies (depth or breadth) and how they relate to the agenda salience of rules the reset and run commands alternatives to rule-based systems case-based reasoning model-based reasoning planning frame problem: actions can have side effects sequencing problem: actions can have preconditions means-end analysis STRIPS: operators with preconditions, add list, and delete list combinatorial explosion problem mini-max game playing game tree alpha-beta pruning static evaluation function "Pierre Menard" and "Funes the Memorious" the problem of meaning importance of abstraction "Understanding Computers and Cognition" blindness representing a situation as facts (objects, properties, relations) the importance of abstraction breakdown source of blindness in symbolic AI programs human vs. computer response to breakdown common sense structural coupling between an organism and an environment understanding natural language the problem of meaning