AI Spring 2010 Lecture 3-5

AI Spring 2010 Lecture 3-5 - Artificial Intelligence...

Info iconThis preview shows pages 1–11. Sign up to view the full content.

View Full Document Right Arrow Icon
Click to edit Master subtitle style Artificial Intelligence Lectures 3-5: Problem Solving and Spring 2010 Instructor: Paul S. Rosenbloom
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
22 This 1.5 Week’s Lectures £ Problem solving £ Search £ Blind search r Search strategies r With repeated states £ Knowledge in Search l How can knowledge help search? l Best-first search and its variants l More on heuristic functions
Background image of page 2
33 Problem Solving in AI £ Fundamentally about achieving a goal l Being rich l Winning a game of chess l Getting to a particular location l Saving people from a burning building l Getting an A in this course £ May take on a variety of variant formulations l Find any state that matches goal l Find all states that match goal l Prove that no state matches goal l Find best state that matches goal l Find best path to a goal state
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
44 Core Activity Determine Action to Perform Next £ May have knowledge that tells you what to do l This is the predominant mode for human experts l They directly “see” what to do next l But will look at knowledge later in course £ Can search over sequences of actions to find one that achieves goal and then pick first action in sequence l Search is fundamentally a response to a lack of knowledge l But it itself requires knowledge in the form of a world model o Could search in world instead of in a model, but this essentially involves making a random next move ± Generally referred to as exploration or online search ± Can lead to learning a world model for later use in internal search
Background image of page 4
55 Problem Formulation £ Given goal, first step is to formulate the problem for search £ First Step: Specify an appropriate world model as a problem space comprising l A set of states representing situations l A set of actions which move among states Then will need to instantiate problem in space
Background image of page 5

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
66 Route Finding in Romania Traveling Salesman Problem is one variation where must visit each city exactly once and want lowest cost path
Background image of page 6
77 Specifying States £ Actions must be specified explicitly, but states may be specified l Explicitly as a list l Implicitly as a cross-product of features l UB: [((B, W) F (P, R, N, B, K, Q)) + Blank]Sq = 1364 (~1071) l Implicitly via an initial state and a set of actions
Background image of page 7

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
88 Instantiating Problem in Space £ Given problem space, second step is to instantiate problem in space l Current/initial state representing starting state for problem solving l Goal state(s) representing situations that meet goal l Can be explicit as a list of one or more states l Can be implicit as a binary test on states ± Checkmate: King under attack and with no ability to move to a safe square, block attack or take attacker
Background image of page 8
99 Eight Puzzle £ Initial State: A configuration of Tiles £ Goal State: Another configuration £ Operators: Slide a tile into space l Or “move” space l Can actually only reach half of all configurations £ Search spaces get dramatically larger as get to fifteen puzzle, twenty four puzzle, etc.
Background image of page 9

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
1010 Formulating the Mutilated Checkerboard
Background image of page 10
Image of page 11
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 03/05/2010 for the course CS 561 taught by Professor Moradi during the Spring '09 term at USC.

Page1 / 115

AI Spring 2010 Lecture 3-5 - Artificial Intelligence...

This preview shows document pages 1 - 11. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online