ShyhKang Jeng

1
Introduction

• Shyh-Kang Jeng
• Department of Electrical Engineering/
• Graduate Institute of Communication Engineering
• National Taiwan University

2
Outlines

• Origin
• Course Outline
• Features
• Intelligence
• Mobility
• Cooperation
• Application Examples
• Some Related Links

3
Origin of Intelligent Agents

• Artificial intelligence
• Internet and Web
• e-Business
• Intelligent software used in Internet
• To provide personalized information
• To perform automated negotiation
• To perform planning and scheduling functions

4
Course Goal

• Give students necessary trainings in order that
  the students can design and implement mobile
  intelligent agents

5
Course Outline

• Advanced Java and Mobility
• Experiments and discussions
• Mobile Agent Systems
• Intelligence
• Problem solving by search
• Knowledge representation and reasoning
• Learning
• Planning
• Multi-agent System
• Literature Reading and Presentation
• Term Project

6
Course Website
http//cc.ee.ntu.edu.tw/skjeng/IntelligentAgent20
03Spring.htm
7
Agent Course Offered by Professor Jane Hsu
http//hugo.csie.ntu.edu.tw/yjhsu/course/u1760/
8
What is an Agent?

• A computational system which
• Is long-lived
• Has goals, sensors, and effectors
• Decides autonomously which actions to take in the
  current situation to maximize progress toward its
  (time-varying) goals

9
Agents
10
What is a Software Agent?

• Particular type of agent, inhabiting computers
  and networks, assisting users with computer-based
  tasks

11
Essence of Software Agents

• Agency
• The degree of autonomy the agent has in
  representing the user to other agents,
  applications, and computer systems
• Intelligence
• The ability of the agent to solve problems
• Mobility
• The capability to move between systems in a
  network

12
Nature of Task Performed

• User agents
• Assist user, know interests/preferences/habits,
  may act on users behalf
• e.g., personal news editor, personal e-shopper,
  personal web guide
• Service agents
• Perform more general tasks in the background
• e.g., web indexing, info retrieval, phone network
  load balancing

13
Agents vs. Expert Systems

• Personalized
• Agents ? different actions
• Expert systems ? same actions
• Active, autonomous
• Agents ? on their own
• Expert systems ? passively answer
• Adaptive
• Agents ? Learn and change
• Expert systems ? Remain fixed

14
Two Problems for Building Software Agents

• Competence
• How does an agent acquire the knowledge it needs?
• Trust
• How can we guarantee the user feels comfortable
  delegating tasks to an agent?

15
User Programmed Agents
16
Knowledge-Based Agents
17
Concrete Architectures for Intelligent Agents

• Logic based agents
• Reactive agents
• Belief-desire-intention agents
• Layered architecture

M. Wooldridge, Intelligent Agents, In G. Weiss,
editor Multiagent Systems, The MIT Press,
April 1999.
18
Logic-Based Architecture

• Internal state is a database of formulae of
  first-order predicate logic
• Examples
• Open(valve221)
• Temperature(reactor4726, 321)
• Pressure(tank776, 28)
• D set of sets of formulae
• D member of D
• A Set of actions

19
Action Selection in Logic Based Architectures

• A action(D D)
• for each a in A
• if (D can derive Do(a)) return a
• for each a in A
• if (D does not derive !Do(a))
• return a
• return null

20
Reactive Architectures

• Decision making is realized through a set of task
  accomplishing behaviours
• Each behavior can be thought as an individual
  action function, which continually takes
  perceptual input and maps it to an action perform
• Each of these behavior modules is intended to
  achieve some particular task

21
Action function for Reactive Architectures

• A action(P p)
• construct set fired as a set of (c, a),
  where c is a set of percepts
• for each (c, a) in fired
• if !( (c, a) exists in fired such that the
  priority of (c, a) is higher than (c, a) )
  return a
• return null

22
Belief-Desire-Intention Architectures Intentions

• Example
• You just leave university with a first degree and
  make a decision about what to do with your life
• Understand options available
• Choose and commit to some options
• Intentions are the chosen options, which
  determines the agents actions

23
Belief-Desire-Intention Architecture
24
Action Selection in BDI Architectures

• A action( P p )
• B brf(B, p)
• D options(D, I)
• I filter(B, D, I)
• return execute(I)

25
Layered Architectures
26
Learning from the User
User
Interacts with
collaborate
Observes And imitates
Application
Interacts with
Agent
27
Conditions to Be Fulfilled by Learning Agents

• Repetition
• The use of the application has to involve a
  substantial amount of repetitive behavior
• Within the actions of one user or among users
• Differentiation
• This repetitive behavior is potentially different
  for different users

28
Sources of Learning

• Look over the shoulder of the user
• Direct and indirect user feedback
• Examples given explicitly by the user
• Advice from agents that assist other users with
  the same task

29
Learning from Other Agents
User
Application
Agent
Agent
Application
User
30
Location of Agents

• Stationary in client
• Stationary in server
• Mobile client-gtserver-gtserver-gt

31
Aglets and Tahiti
32
NTUEE XMAS
???, ???????????????????,
?????????, 1999.
33
Mobility of Agents

• NOT a necessary characteristic for some program
  to be an agent
• Why move agents around?
• Reduce network traffic
• Share load among machines
• Go to the data if the data cant come to you
• User may have only infrequent connection to
  network
• Seven good reasons for using mobile agents, in
  D. B. Lange/M. Oshima, Programming and Deploying
  Java Mobile Agents with Aglets, pp. 3-5.

34
Security Problems

• Agent protection
• Remote host threatens agent
• Agent threatens another agent
• Unauthorized third parties threatens agent
• Host protection
• Incoming agent threatens host
• Unauthorized third parties threaten host
• Network protection
• Incoming agent threatens the network

35
Common Language, Common Ontology

• Homogeneous agents ? no problem
• Heterogeneous agents ? user standard like KQML,
  etc.

36
Modeling Other Agents

• Logical approach
• Beliefs, desires, intentions (e.g. Shohams work
  at Stanford)
• Pragmatic approach
• Agents keeping a trust level of other agents for
  a set of problems (e.g., Maes et al work at MIT)

37
Agent Maxims
38
Eager Assistant Agent
Memory of examples
Situation1? action1
New situation
Situation2? action2
Situation3? action3
Predicted action, Confidence level
SituationN? actionN
39
Details of One Example

• Situation
• Type new message received
• Sender nicholas_at_media.mit.edu
• Date 3/10/95 1604
• Topic meet to discuss funding
• Receiver pattie_at_media.mit.edu
• Body ltgt
• Keywordsltgt
• Action taken
• Message read first out of 20
• Message read first time

40
Prediction and Confidence Level Computation

• Prediction
• Compute k nearest situations and distances ds
• Compute score for every action S1/ds
• Pick action with highest score
• Confidence level
• Higher if ds smaller
• Higher if less disagreement
• Higher if more examples in memory

41
Using the Prediction

• Agent operates directly on confidence level for
  every action that can be automated

1
Do-it threshold
Tell-me threshold
0
42
Maxims Agent States
43
Multi-agent Collaboration

• If agents confidence lt tell-me threshold
• Agent contacts other agents via bboard
• Agent sends details of situation
• Some agents respond with Pi and Ci
• Agents computes score for every action SCiTi,s
• Agent picks action with highest score
• Agent updates trust levels Ti,s

44
Meeting Scheduling Agent
45
Results and Confidence Level
46
Agent NewT
47
Feature-Based Filtering Analyzing Documents for
Relevant Keywords

• SMART algorithm (Salton, Cornell, 69)
• Remove stop words
• Stem other words
• Compute ratio (frequency of word in this
  document)/(average frequency of word in all
  documents)
• Pick n words with highest ratios as the
  representation of the document

48
Feature-Based Filtering Updating User Profile
Based on One More Datapoint Document

• If user likes document, then increases weights of
  relevant keywords of the document in the profile
• If user dislikes document, then decreases weights
  of relevant keywords of the document in the
  profile

49
Feature-Based Filtering Filtering Based on User
Profile

• Extract relevant keywords of document
• Compare those with keywords in users profile
• Compute score of document S weightratio of
  occurrence for all keywords in the profile
• If score above threshold then present to user

50
Typical Results of Agent Ringo
51
Automated Collaborative Filtering (ACF)
User-2
User-1
User-k
User-n
5
7
6
v1n
Item-1
1
v2n
2
Item-2
7
v3n
6
/
3
Item-k
5
vln
4
6
Item-l
vmn
Item-m
/
2
3
52
Collaborative vs. Feature Based Techniques

• FBF techniques attempt analysis of object
  content ACF does not
• Can be applied to areas like music where content
  not easily analyzed

53
Comparison (cont.)

• FBF techniques raise why questions ACF dodges
  this bullet
• Assume that comparable people like comparable
  things for comparable reasons
• FBF techniques require potentially unbounded
  real-world knowledge ACF uses the knowledge of
  other people
• Agent is actually built with no AI at all!

54
Dispatching Learning Agent for TSMC
???, ??????????????????????,
?????????, 2002
55
Letiziahttp//lieber.www.media.mit.edu/people/lie
ber/Lieberary/Letizia/Letizia.html/
56
Letizia

• Agent for assisting web browsing
• Human browsing usually depth-first. Agent can
  augment this by doing breadth-first browse
• When you look at a page, agent
• Retrieves nearby links
• Analyze pages for prominent keywords
• Pre-fetches interesting docs to 2nd window

57
Yentahttp//foner.www.media.mit.edu/people/foner/
yenta-from-old-pages.html
58
Yenta

• System for connecting/introducing people based on
  interests and activities
• Matchmaking (1-to-1)
• Coalition-building (n-to-n)
• Technical goals
• Distributed avoid single failure bottlenecks
• Use cryptography careful design to protect user
  privacy

59
Kasbahhttp//xenia.media.mit.edu/nelson/research
/kasbah/node3.html
60
Kasbah

• Kasbah intelligent classified ad.
• The ad is an agent. Given key data it
• Searches out compatible agents in a centralized
  marketplace (sellers look for buyers/bidders
  vice versa).
• Conducts business for users.
• Completes transaction, if authorized.

61
A Search for Intelligent Agents via Yahoo!
62
UMBC Agent Webhttp//agents.umbc.edu/
63
BottomDollarhttp//www.bottomdollar.com/index.php
/ut8c7015a036191a61
64
Personal Travel Assistant http//www.cs.sfu.ca/la
md/personal/cmpt882/PTA.html/
65
SpiderHunter.comhttp//www.spiderhunter.com/
66
Intelligent Agent Systems Labhttp//iasl.iis.sini
ca.edu.tw/iaslindex.html
67
MIT Media Lab Software Agents Grouphttp//agents
.media.mit.edu/index.html
68
Pattie Maeshttp//pattie.www.media.mit.edu/people
/pattie/
69
CMU Intelligent Software Agent Labhttp//www-2.cs
.cmu.edu/softagents/
70
AgentCitieshttp//www.agentcities.org/
71
Semantic Webhttp//www.sciam.com/article.cfm?arti
cleID00048144-10D2-1C70-84A9809EC588EF21
72
Semantic Web Agents