introduction to agents and multi-agent systems (lecture slides)

67
D. Monett Europe Week 2014, University of Hertfordshire, Hatfield Introduction to Agents and Multi-agent systems Prof. Dr. Dagmar Monett Díaz Computer Science Dept. Faculty of Cooperative Studies Berlin School of Economics and Law [email protected] Europe Week, 3 rd 7 th March 2014

Upload: dagmar-monett

Post on 15-Jan-2015

543 views

Category:

Education


1 download

DESCRIPTION

Online lecture at the School of Computer Science, University of Hertfordshire, Hatfield, UK, as part of the 10th Europe Week from 3rd to 7th March 2014.

TRANSCRIPT

Page 1: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Introduction to Agents and

Multi-agent systems

Prof. Dr. Dagmar Monett Díaz Computer Science Dept.

Faculty of Cooperative Studies

Berlin School of Economics and Law

[email protected]

Europe Week, 3rd – 7th March 2014

Page 2: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield 2

Two “intelligent” agents…

Page 3: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Dilbert

Scott Adams

At http://dilbert.com/strips/comic/1991-02-08/

(Educational/Classroom usage permission is granted by Universal Uclick. All Rights Reserved)

“Intelligent” travel agent #1

3

Page 4: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Dilbert

Scott Adams

At http://dilbert.com/strips/comic/1996-12-23/

(Educational/Classroom usage permission is granted by Universal Uclick. All Rights Reserved)

“Intelligent” travel agent #2

4

Page 5: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield 5

Another kind of “intelligent” agent

Page 6: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Travel agent: simple case

"Stimulus-response„ search engine

Customer: Agent:

Specifies request with preferences

[Fill out form]

Specifies answer

[Show matching

offers]

6

Page 7: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Travel agent: more complex

Customer: Agent:

I want to go on vacation

Wonderful!

Do you like

swimming?

7

Page 8: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Travel agent: more complex

Customer: Agent:

Yes, with good friends on a white beach. And I like sports.

Wonderful!

And in the evening?

8

Page 9: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Travel agent: more complex

Customer: Agent:

Good entertainment, exclusive bars, etc.

Sounds fantastic. Is this what you are looking for? [Present an offer]

9

Page 10: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Travel agent: more complex

Customer: Agent:

Really fantastic but over my budget. I‘d prefer something less exclusive...

Let's see… How about this? [Present a new offer]

10

Page 11: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield 11

What would that agent need for the

dialog?

Page 12: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Travel agent: more complex

12

Such a travel agent would need its knowledge to be

dynamic:

Page 13: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Travel agent: more complex

13

Such a travel agent would need its knowledge to be

dynamic:

Dialog history

Page 14: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Travel agent: more complex

14

Such a travel agent would need its knowledge to be

dynamic:

Dialog history

(Hypothetical) model of the customer’s and own

• desires, intentions

• preferences, opinions

Page 15: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Travel agent: more complex

15

Such a travel agent would need its knowledge to be

dynamic:

Dialog history

(Hypothetical) model of the customer’s and own

• desires, intentions

• preferences, opinions

(Flexible) plan for

• exploring the customer’s desires and intentions

• profitable offers

Page 16: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield 16

Agenda

Page 17: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield 17

Agenda

Where does the major content come from?

What is an agent? What is a multi-agent system?

Agent types

Agent properties

Design of intelligent agents

Implementing practical reasoning agents

Further reading, sources of inspiration, and more…

Page 18: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield 18

©

Page 19: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

An Introduction to

MultiAgent Systems

Michael Wooldridge

2nd Edition, 484 pp.

John Wiley & Sons, May 2009

ISBN-13: 978-0470519462

With materials available at

http://www.csc.liv.ac.uk/~mjw/pub

s/imas/IMAS2e.html

What I also use in my lectures at the HWR…

19

Page 20: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Artificial Intelligence: A

Modern Approach

Stuart Russell and Peter Norvig

3rd Edition, 1152 pp.

Prentice Hall, December 2009

ISBN-13: 978-0136042594

With materials available at

http://aima.cs.berkeley.edu/

What I also use in my lectures at the HWR…

20

“The AI Bible”

Page 21: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield 21

What is an agent?

Page 22: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Agents, a definition

22

According to Michael Wooldridge:

“An agent is a computer system that is

capable of independent action on behalf of

its user or owner (figuring out what needs to

be done to satisfy design objectives, rather

than constantly being told)”.

Page 23: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Agents, a definition

23

According to Michael Wooldridge:

“The main point about agents is they are

autonomous: capable of acting

independently, exhibiting control over their

internal state.

Thus: an agent is a computer system

capable of autonomous action in some

environment in order to meet its design

objectives”.

Page 24: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Agents, a definition

24

According to Russell and Norvig:

“An agent is anything that can be viewed as

perceiving its environment through sensors

and acting upon that environment through

actuators”.

Page 25: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield 25

Agents

Agents communicate with the environment

through Sensors and Actuators

Page 26: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield 26

What is a multi-agent system?

Page 27: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Dilbert

Scott Adams

At http://dilbert.com/strips/comic/1998-08-29/

(Educational/Classroom usage permission is granted by Universal Uclick. All Rights Reserved)

Agent in a group

27

Page 28: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Multi-agent systems, a definition

28

According to Michael Wooldridge:

“A multi-agent system is one that consists of

a number of agents, which interact with one-

another”.

Page 29: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Multi-agent systems, a definition

29

According to Michael Wooldridge:

“A multi-agent system is one that consists of

a number of agents, which interact with one-

another”.

In the most general case, agents will be acting on

behalf of users with different goals and motivations

To successfully interact, they will require the ability

to cooperate, coordinate, and negotiate with each

other, much as people do

Page 30: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield 30

Agent types

Page 31: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Agent types

Four basic types in order of increasing

generality:

Simple reflex agents

Model-based reflex agents

Goal-based agents

Utility-based agents

31

Page 32: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield 32

Simple reflex agents

Page 33: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield 33

Model-based reflex agents

Page 34: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield 34

Goal-based agents

Page 35: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield 35

Utility-based agents

Page 36: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield 36

Learning agents

Page 37: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Dilbert

Scott Adams

At http://dilbert.com/strips/comic/2004-04-19/

(Educational/Classroom usage permission is granted by Universal Uclick. All Rights Reserved)

A real state agent

37

Page 38: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield 38

Agent properties

Page 39: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Intelligent agents are…

reactive

pro-active

social

rational

benevolent

adaptable

truthful

autonomous

mobile

A reactive system is one that

maintains an ongoing interaction

with its environment, and responds

to changes that occur in it (in time

for the response to be useful).

39

Page 40: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Intelligent agents are…

reactive

pro-active

social

rational

benevolent

adaptable

truthful

autonomous

mobile

Pro-activeness = generating and

attempting to achieve goals; not

driven solely by events; taking the

initiative; recognizing opportunities.

40

Page 41: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Intelligent agents are…

reactive

pro-active

social

rational

benevolent

adaptable

truthful

autonomous

mobile

Social ability in agents is the ability

to interact with other agents (and

possibly humans) via some kind of

agent-communication language, and

perhaps cooperate with others.

41

Page 42: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Intelligent agents are…

reactive

pro-active

social

rational

benevolent

adaptable

truthful

autonomous

mobile

An agent should strive to do the

right thing, based on what it can

perceive and the actions it can

perform. The right action is the one

that will cause the agent to be most

successful.

An agent will act in order to achieve

its goals, and will not act in such a

way as to prevent its goals being

achieved – at least insofar as its

beliefs permit

42

Page 43: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Intelligent agents are…

reactive

pro-active

social

rational

benevolent

adaptable

truthful

autonomous

mobile

Agents do not have conflicting

goals. Every agent will therefore

always try to do what is asked of it.

43

Page 44: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Intelligent agents are…

reactive

pro-active

social

rational

benevolent

adaptable

truthful

autonomous

mobile

Agents improve performance over

time. They can learn.

44

Page 45: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Intelligent agents are…

reactive

pro-active

social

rational

benevolent

adaptable

truthful

autonomous

mobile

An agent will not knowingly

communicate false information

45

Page 46: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Intelligent agents are…

reactive

pro-active

social

rational

benevolent

adaptable

truthful

autonomous

mobile

An agent is autonomous if its

behaviour is determined by its own

experience (with ability

to learn and adapt)

46

Page 47: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Intelligent agents are…

reactive

pro-active

social

rational

benevolent

adaptable

truthful

autonomous

mobile

Mobility: the ability of an agent to

move around an electronic network

47

Page 48: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield 48

Design of intelligent agents.

Examples

Page 49: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Settings

49

Performance measure

Environment

Sensors

Actuators

Page 50: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Settings

50

Performance measure

Environment

Sensors

Actuators

Designing an automated taxi driver:

Safe, fast, legal, comfortable trip, maximize profits

Roads, other traffic, pedestrians, customers

Cameras, speedometer, GPS, engine sensors, keyboard, etc.

Steering wheel, accelerator, brake, signals

Page 51: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Settings

51

Performance measure

Environment

Sensors

Actuators

Designing a medical diagnosis system:

Healthy patient, minimize costs, lawsuits

Patient, hospital, staff

Keyboard (entry of symptoms, findings, patient's answers)

Screen display (questions, tests, diagnoses, treatments, referrals)

Page 52: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Settings

52

Performance measure

Environment

Sensors

Actuators

Designing an interactive English tutor:

Maximize student's score on test

Set of students

Keyboard

Screen display (exercises, suggestions, corrections)

Page 53: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield 53

Active learning exercise:

“How to implement

reactive agents?

Discuss it with your classmates!”

Image © renjith krishnan at http://www.freedigitalphotos.net/

Page 54: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield 54

Implementing deliberative agents.

First steps.

Page 55: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Agent Control Loop Version 1 while true observe the world; update internal world model; deliberate about what intention to achieve next; use means-ends reasoning to get a plan for the intention; execute the plan end while

First pass

Page 56: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

More formally…

Page 57: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

With deliberation…

[…which can be further extended to consider practical reasoning agents…]

Page 58: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Dilbert

Scott Adams

At http://dilbert.com/strips/comic/2004-04-23/

(Educational/Classroom usage permission is granted by Universal Uclick. All Rights Reserved)

The real state agent (cont.)

58

Page 59: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield 59

Homework:

“Search for real-world applications

that use intelligent agents and

describe their tasks and

functioning!”

Image © renjith krishnan at http://www.freedigitalphotos.net/

Page 60: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield 60

Assessment

Image © renjith krishnan at http://www.freedigitalphotos.net/

Page 61: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Question

61

(Taken from a public sample exam © Wooldridge)

Briefly define and explain, with examples where

appropriate, THREE properties you would

expect an intelligent agent to exhibit!

PLEASE ANSWER AT:

https://docs.google.com/forms/d/1K0RZur9bDtZs0R0k74GKn7Va34ALPYtXxXUFJH51_1k/viewform

Page 62: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield 62

Last active learning exercise

Image © renjith krishnan at http://www.freedigitalphotos.net/

Page 63: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield 63

The content

Page 64: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield 64

References

Page 65: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Others…

65

Page 66: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield

Further reading and sites…

IFAAMAS – International Foundation for

Autonomous Agents and Multiagent Systems

http://ifaamas.org/

Journal Autonomous Agents and Multi-Agent

Systems

http://link.springer.com/journal/10458

13th International Conference on Autonomous

Agents ans Multiagent Systems (AAMAS’14), Paris,

France

http://aamas2014.lip6.fr/

66

Page 67: Introduction to Agents and Multi-agent Systems (lecture slides)

D. Monett – Europe Week 2014, University of Hertfordshire, Hatfield 67

Slides of the talk per request:

[email protected]

Prof. Dr. Dagmar Monett Díaz

monettdiaz

@dmonett

http://monettdiaz.com