COGS 100 Lecture Notes - Lecture 13: General Problem Solver, Forward Chaining, Knowledge Representation And Reasoning
Document Summary
Cogs 100 - introduction to cognitive science - lecture 13: representations: rules. Rules: mental representation of the form if (condition) then (action) Logic can be used to find the truth value of the condition. A form of knowledge representation in ai; used in production systems where using forward chaining, matching rules for a given state are discovered and the corresponding action is executed to move the system toward a goal state. Newells and simon, general problem solver, 1950s 1960s. Soar: newell, laird, etc. and act-r, 1983 now, john anderson, cmu have evolved as cognitive models. Forward chaining: string rules together, matched against working memory of current state. Backward chaining: work back from goal to start. Generalize to form new rules induction. Specialize to form more specific rules deduction. Abduction to form explanations: if cause then effect, effect, so, maybe cause. It is difficult to express probability using simple rules (the truth value for condition is either true or false)