Presentation is loading. Please wait.

Presentation is loading. Please wait.

Agents the AI metaphor. D Goforth - COSC 4117, fall 20062 The agent model  agents include all aspects of AI in one object-oriented organizing model:

Similar presentations


Presentation on theme: "Agents the AI metaphor. D Goforth - COSC 4117, fall 20062 The agent model  agents include all aspects of AI in one object-oriented organizing model:"— Presentation transcript:

1 Agents the AI metaphor

2 D Goforth - COSC 4117, fall 20062 The agent model  agents include all aspects of AI in one object-oriented organizing model: e n v i r o n m e n t AGENT actperceive purpose

3 D Goforth - COSC 4117, fall 20063 Agents and decentralization Mars Rover direct control  agent 

4 D Goforth - COSC 4117, fall 20064 Purpose  Why do agents act? goals  internal (state of agent’s structure, eg survive)  external (state of environment, eg clean up dirt)  How to measure success? compare actual results to goals R&N ‘performance measure’

5 D Goforth - COSC 4117, fall 20065 Performance measure  external to agent (like javadoc specification)  ideal that cannot always be achieved completely (unlike javadoc specs)  Agent success (‘rationality’) is evaluated based on performance measure AND percepts, possible actions, experience (like an athlete)

6 D Goforth - COSC 4117, fall 20066 Factors in rationality  performance measure – goals may be in conflict – can’t all be achieved  perceptions – agent may not have all the facts  actions available  experience – agent may not yet have accumulated all available relevant data

7 D Goforth - COSC 4117, fall 20067 Agents are not just methods  actual outcome of actions are not known 100%  algorithms are not complete solutions – agents should be partly autonomous  learn from experience gather data about environment respond better to same perceptions

8 D Goforth - COSC 4117, fall 20068 The agent model  agents include all aspects of AI in one organizing model: e n v i r o n m e n t AGENT actperceive purpose

9 D Goforth - COSC 4117, fall 20069 Example CASH REGISTER AS AGENT Goals: get payment for items, update inventory, accumulate payments Perceive bar code Know price lists Understand finding prices, names from bar code Understand accumulating bill Act to send price, code to accounting, send inventory change to db. Act to display item name, price, and running total Perceive signal for no-more-items Act to request payment Perceive payment …

10 D Goforth - COSC 4117, fall 200610 Environments  real or virtual  may contain other agents factors relevant to the agent are called the state of the environment perceptions give agent information about the state actions of agent change the state

11 D Goforth - COSC 4117, fall 200611 Categorizing Environments (R&N p40-44) 1.fully or partly observable – perception of state of the environment game examples: chess, bridge, Myst

12 D Goforth - COSC 4117, fall 200612 Categorizing Environments (R&N p40-44) 2.actions are predictable – deterministic vs. stochastic vs. strategic game examples: chess, Monopoly, solitaire yogo peg game, solitaire card game

13 D Goforth - COSC 4117, fall 200613 Categorizing Environments (R&N p40-44) 3.episodic vs. sequential – actions are based on how many previous perceptions and actions? game examples: chess, paper-scissors- rock, bridge trick, bridge hand

14 D Goforth - COSC 4117, fall 200614 Categorizing Environments (R&N p40-44) 4.real-time vs event driven- (static vs dynamic) agent and environment are sequential or co-routines game examples: chess, tetris

15 D Goforth - COSC 4117, fall 200615 Categorizing Environments (R&N p40-44) 5.discrete vs. continuous environment, perception, action game examples: chess, tetris, driving simulator

16 D Goforth - COSC 4117, fall 200616 Categorizing Environments (R&N p40-44) 6.number of agents – 1 or more competitive, cooperative, codependent, interfering, communicating (info separate from perceptions) game examples: solitaires, chess, bridge, futures, tetris, driving simulator(s), role playing games

17 D Goforth - COSC 4117, fall 200617 Categorizing King’s Court: 1.fully / partly observable 2.deterministic / stochastic 3.sequential / episodic 4.static / dynamic 5.discrete / continuous 6.single- / multi-agent

18 D Goforth - COSC 4117, fall 200618 Agent Structure  agent program is ‘episodic’ – receives percepts and produces actions (parameters and return values) BUT internal state of agent can evolve sequentially – agent may be in a different state after episode than before

19 D Goforth - COSC 4117, fall 200619 Agent Structure 1.Table-Driven (p.45) single perception look-up (HUGE table) perception sequence look-up (HUGER table) example game: tic-tac-toe perfect solution but intractible

20 D Goforth - COSC 4117, fall 200620 Table-driven agents (revised from R&N) LOOK-UP TABLE Keyvalue Percept1action1 Percept2action2 … KNOWLEDGE

21 D Goforth - COSC 4117, fall 200621 Agent Structure 2.Simple reflex (p.46) based on current perception only i.e., no instance variables in the agent object; no state ‘condition-action’ rules (if then else algorithm)

22 D Goforth - COSC 4117, fall 200622 Simple reflex agents – R&N

23 D Goforth - COSC 4117, fall 200623 Agent Structure 3.Model-based reflex (p.48) uses percepts to build internal model of environment - internal state is ‘memory’ of environment algorithm based on percepts and internal state

24 D Goforth - COSC 4117, fall 200624 Model-based reflex agents – R&N

25 D Goforth - COSC 4117, fall 200625 Agent Structure 4.Goal-based (p.49) internal state representing environment PLUS goals expressed in terms of environment and/or agent states NOT REFLEX; ‘tries’ actions internally and tests results against goals

26 Goal-based agents – R&N

27 D Goforth - COSC 4117, fall 200627 Agent Structure 5.Utility-based (p.50) internal state representing environment PLUS goals expressed in terms of environment and/or agent states PLUS performance measure rationality ‘tries’ actions internally and tests results against goals AND performance measure

28 D Goforth - COSC 4117, fall 200628 Utility-based agents – R&N

29 D Goforth - COSC 4117, fall 200629 Agent Structure 6.Learning(p.50) extra component to evaluate performance and change program (if necessary) to act differently in same state many kinds of learning agents

30 Learning agents – R&N

31 D Goforth - COSC 4117, fall 200631 Agent Structures 1.Table driven 2.Simple reflex 3.Model-based reflex 4.Goal-based 5.Utility-based 6.Learning

32 D Goforth - COSC 4117, fall 200632 Example CASH REGISTER AS AGENT: Goals: get payment for items, update inventory, accumulate payments Perceive bar code Know price lists Understand finding prices, names from bar code Understand accumulating bill Act to send price, code to accounting, send inventory change to db. Act to display item name, price, and running total Act to request payment Perceive payment …


Download ppt "Agents the AI metaphor. D Goforth - COSC 4117, fall 20062 The agent model  agents include all aspects of AI in one object-oriented organizing model:"

Similar presentations


Ads by Google