31
1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

Embed Size (px)

Citation preview

Page 1: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

1

An Introduction to Agent Technologies

Peter Wurman, NCSUYelena Yesha, UMBCOlga Streltchenko, UMBC

Page 2: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

2

Presentation Overview

Working definition of an agentAgent characteristics and propertiesAgent societiesExamples

Page 3: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

3

Working definition of an agent

“Agents are active, persistent (software) components that perceive, reason, act, and communicate” Huhns and Singh, 1998

“An agent is an entity whose state is viewed as consisting of mental components such as beliefs, capabilities, choices, and commitments. [sic] In this view, therefore, agenthood is in the mind of the programmer.” Shoham, 1993

Page 4: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

4

Agent Program

Inputs = observations Observations: states of the agent’s domain or

environment

Outputs = actions Actions: Speak, Search, Move, Bid

Agent

( o1, o2, … ) ( a1, a2, … )

Page 5: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

5

Agents Environments

An agent must have a model of its domain and a model of other agents that it communicates with.

Properties of agents’ environment: Observable Dynamic Discrete

Page 6: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

6

Basic Characteristics

Delegation abilities: The owner or user of an agent delegates a task to the agent and the agent autonomously performs the task on behalf of the user. An agent can decompose and/or delegate

the task to other agents; Once the task is complete the agent may

need to report to the user/agent issuing the task.

Page 7: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

7

Basic Characteristics (cont’d)

Agent communication languages and protocols: information exchange with other agents establishes a need for expressive communication and negotiation language. KQML (Knowledge Query and Manipulation

Language);Used to allow information agents to assert interests

in information services, advertise their own services, and explicitly delegate tasks and requests for assistance from other agents.

Can be used for developing a variety of inter-agent communication protocols that enable information agents to collectively cooperate.

Page 8: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

8

Basic Characteristics (cont’d)

Self-representation abilities: the ability to express business and system aspects of its functionality, combine them into an application or implementation. Self-describing, dynamic reconfigurable

agents;Facilitate composition (specification and

implementation) of large-scale (distributed) applications.

Page 9: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

9

Agent Mental State (BDI)

Beliefs–knowledge about the world and the effects the agent’s actions have on the world.

Desires–preferences over possible states of the world (goals).

Intentions–internal commitments made to achieve certain world states.

Page 10: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

10

Example: Trading Agent

User preferences

Auctionrules

Model of other market participants

Strategysynthesizer

Biddingstrategy

Market Info

Bids

Page 11: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

11

Example: Trading Agent

Beliefs: auction rules, model of market

Desires: user preferencesIntentions: objects it has decided to

buy/sellCapabilities: place new bidsObservations: market information

Page 12: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

12

Reactive Agent

Lookup table maps each observation, or series of observations, to an action

an = f(on), oran = f(o1,…, on)

Fast Inflexible Intractable for nontrivial domains

Page 13: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

13

Rational Agents

Decision theoretic (economic) Agent makes optimal decisions given its

beliefs, goals, and intentions.Logical

Agent makes decisions that are consistent with its beliefs, goals, and intention.

Page 14: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

14

Boundedly Rational Agents

Agent makes optimal decisions given its beliefs, goals, intentions, and the limits of its computational abilities.

Page 15: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

15

Learning Agent

An agent that updates its beliefs based on its observations

What can we learn? Model of the world New capabilities Effects of our actions

Page 16: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

16

Learning Agent

Learning task: Learn , where sn+1 = (sn, an)

Types of learning Supervised Reinforcement Unsupervised

Page 17: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

17

Autonomy

Agent autonomy, with respect to User = execution autonomy Other agents = social autonomy

Designer autonomy, with respect to Communication = interface autonomy Architecture = design autonomy Utility function = preference autonomy

Page 18: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

18

Belief Representation

Knowledge level The Wolfline runs from HC to CC

Logical level (declarative) Connects(Wolfline,HC,CC)

Implementation level (procedural)public class Bus{

public string start = “HC”;public string end = “CC”;}

Page 19: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

19

Benefits of Declaritivism

ModularitySemanticsInspectabilityLearnabilityProgrammability

Page 20: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

20

Other Properties of Agents

Lifespan: transient to long-livedModeling: of itself, the world, and

other agentsMobility: stationary or mobileMemory: non to perfect recall

Page 21: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

21

Agent Societies

Software infrastructureSocial organization

Page 22: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

22

Software Infrastructure

Communication ChannelsCommon OntologiesService agents (i.e. directory agent)

Page 23: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

23

Communication

ACL = agent communication language

Example: KQML Content messages: tell, query, reply, etc.

Flow control: next, etc.Generally do not prescribe semantics

Page 24: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

24

Ontologies

Define the semantics of communication

Notoriously difficult Employee = everyone on payroll Employee = everyone receiving benefits

Page 25: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

25

Social Organization

Homogeneous or heterogeneousSelf-interest v.s. cooperativeSocial control structureSystem evaluation

Page 26: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

26

Emergent Behavior

Agents act With limited, local knowledge In self interest

System behaves In globally desirable manner Without central control

Adam Smith’s “invisible hand”

Page 27: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

27

E-commerce example

Trading agents, again Heterogeneous Self-interested Mediated Game-theoretic evaluations

Page 28: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

28

Legacy System Example

Agent

DB1 DB2 DB3

AgentAgent

UserAgent

Page 29: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

29

Personal Assistants

Agents that support a user’s task Example (weak)

Dialogue basedAnthropoidCooperative ?But has no goals of its own

Page 30: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

30

Personal Assistants

Example: smart calendar/datebook that could Negotiate appointments for me Actively keep track of my contacts by

searching the web Learn priorities for my mail

Page 31: 1 An Introduction to Agent Technologies Peter Wurman, NCSU Yelena Yesha, UMBC Olga Streltchenko, UMBC

31

Conclusion

Agenthood is a convenient descriptionAgents are described by beliefs, desires,

and intentionsAgents select actions based on

observationsCooperating agents are a form of

distributed computationSelf-interested agents can generate

desirable emergent behavior