Decision Making and AI

8/2/2019 Decision Making and AI

1/38

McGraw-Hill/Irwin 2009 The McGraw-Hill Companies, All Rights

Chapter 4DECISION SUPPORT AND ARTIFICIAL

INTELLIGENCE

Brainpower for Your Business

8/2/2019 Decision Making and AI

2/38

4-2

STUDENT LEARNINGOUTCOMES

1. Compare and contrast decisionsupport systems and geographicinformation systems.

2. Define expert systems and describethe types of problem to which theyare applicable.

3. Define neural networks and fuzzy

logic and the use of these AI tools.

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

3/38

4-3

STUDENT LEARNINGOUTCOMES

4. Define genetic algorithms and list theconcepts on which they are basedand the types of problems they solve.

5. Describe the four types of agent-based technologies.

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

4/38

4-4

VISUALIZING INFORMATION INMAP FORM FOR DECISION MAKING

o Geographic information systems(GISs) allow you to see informationspatially, or in map form.

o Researchers and scientists used aGIS to map the location of all thedebris from the shuttle Columbia

o The city of Chattanooga uses a GIS to

map the location of its 6,000 trees tohelp develop a maintenance schedule

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

5/384-5

VISUALIZING INFORMATION IN MAPFORM FOR DECISION MAKING

o The city of Richmond, VA, used a GISto optimize its 2,500 bus stoplocations in its public transportation

systemo Sometimes, a picture is worth a

thousand words

o Recall from Chapter 1, the form of

information often defines its quality

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

6/384-6

VISUALIZING INFORMATION IN MAPFORM FOR DECISION MAKING

1. Do you use Web-based mapservices to get directions andfind the location of buildings? If

so, why?2. In what ways could real estate

agents take advantage of the

features of a GIS?3. How could GIS software benefit

a bank wanting to determine theoptimal placements for ATMs?

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

7/384-7

INTRODUCTION

o Phases of decision making

1. Intelligence find or recognize aproblem, need, or opportunity

2. Design consider possible ways ofsolving the problem

3. Choice weigh the merits of eachsolution

4. Implementation carry out the solution

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

8/384-8

Four Phases of Decision Making

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

9/384-9

Types of Decisions You Face

o Structured decision processing a certaininformation in a specified way so that youwill always get the right answer

o Nonstructured decision one for which

there may be several right answers,without a sure way to get the right answer

o Recurring decision happens repeatedly

o Nonrecurring (ad hoc) decision oneyou make infrequently

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

10/384-10

Types of Decisions You Face

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

EASIEST

MOST

DIFFICULT

8/2/2019 Decision Making and AI

11/384-11

CHAPTER ORGANIZATION

1. Decision Support Systems

Learning outcome #1

2. Geographic Information Systems

Learning outcome #13. Expert Systems

Learning outcome #2

4. Neural Networks and fuzzy Logic

Learning outcome #3

8/2/2019 Decision Making and AI

12/384-12

CHAPTER ORGANIZATION

5. Genetic Algorithms

Learning outcome #4

6. Intelligent Agents

Learning outcome #5

8/2/2019 Decision Making and AI

13/384-13

DECISION SUPPORT SYSTEMS

o Decision support system (DSS) ahighly flexible and interactive systemthat is designed to support decision

making when the problem is notstructured

o Decision support systems help youanalyze, but you must know how to

solve the problem, and how to use theresults of the analysis

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

14/384-14

Alliance between You and a DSS

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

15/384-15

Components of a DSS

o Model management componentconsists of both the DSS models andthe model management system

o Data management componentstores and maintains the informationthat you want your DSS to use

o User interface management

component allows you tocommunicate with the DSS

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

16/384-16

Components of a DSS

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

17/38

4-17

GEOGRAPHICINFORMATION SYSTEMS

o Geographic information system(GIS) DSS designed specifically toanalyze spatial information

o Spatial information is any informationin map form

o Businesses use GIS software toanalyze information, generate

business intelligence, and makedecisions

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

18/38

4-18

Zillow GIS Software for Denver

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

19/38

4-19

ARTIFICIAL INTELLIGENCE

o DSSs and GISs support decision making;you are still completely in charge

o Artificial intelligence, the science ofmaking machines imitate human thinking

and behavior, can replace humandecision making in some instances

Expert systems

Neural networks (and fuzzy logic)

Genetic algorithms

Intelligent agents (or agent-basedtechnologies)

8/2/2019 Decision Making and AI

20/38

4-20

EXPERT SYSTEMS

o Expert (knowledge-based) systeman artificial intelligence system thatapplies reasoning capabilities to reacha conclusion

o Used for

Diagnostic problems (whats wrong?)

Prescriptive problems (what to do?)

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

21/38

4-21

Traffic Light Expert System

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

22/38

4-22

What Expert Systems Canand Cant Do

o An expert system can Reduce errors

Improve customer service

Reduce cost

o An expert system cant

Use common sense

Automate all processes

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

23/38

4-23

NEURAL NETWORKS ANDFUZZY LOGIC

o Neural network (artificial neuralnetwork or ANN) an artificialintelligence system that is capable offinding and differentiating patterns

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

24/38

4-24

Neural Networks Can

o Learn and adjust to newcircumstances on their own

o Take part in massive parallel

processingo Function without complete information

o Cope with huge volumes ofinformation

o Analyze nonlinear relationships

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

25/38

4-25

Fuzzy Logic

o Fuzzy logic a mathematical methodof handling imprecise or subjectiveinformation

o Used to make ambiguous informationsuch as short usable in computer

systems

o Applications

Googles search engine

Washing machines

Antilock breaks

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

26/38

4-26

GENETIC ALGORITHMS

o Genetic algorithm an artificialintelligence system that mimics theevolutionary, survival-of-the-fittestprocess to generate increasinglybetter solutions to a problem

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

27/38

4-27

Evolutionary Principles ofGenetic Algorithms

1. Selection or survival of the fittestor giving preference to betteroutcomes

2. Crossover combining portions ofgood outcomes to create even betteroutcomes

3. Mutation randomly trying

combinations and evaluating thesuccess of each

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

28/38

4-28

Genetic Algorithms Can

o Take thousands or even millions ofpossible solutions and combine andrecombine them until it finds theoptimal solution

o Work in environments where no modelof how to find the right solution exists

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

29/38

4-29

INTELLIGENT AGENTS

o Intelligent agent software thatassists you, or acts on your behalf, inperforming repetitive computer-relatedtasks

o Types

Information agents

Monitoring-and-surveillance or predictive

agents Data-mining agents

User or personal agents

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

30/38

4-30

Information Agents

o Information Agents intelligentagents that search for information ofsome kind and bring it back

o Ex: Buyer agentor shopping botan intelligent agent on a Web site thathelps you, the customer, find productsand services you want

McGraw-Hill

2007 The McGraw-Hill Companies Inc All rights reserved

M it i d

8/2/2019 Decision Making and AI

31/38

4-31

Monitoring-and-Surveillance Agents

Monitoring-and-surveillance(predictive) agents intelligentagents that constantly observe andreport on some entity of interest, anetwork, or manufacturing equipment,for example

McGraw-Hill

2007 The McGraw-Hill Companies Inc All rights reserved

8/2/2019 Decision Making and AI

32/38

4-32

Data-Mining Agents

Data-mining agent operates in adata warehouse discoveringinformation

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

33/38

4-33

User Agents

o Useror personal agent intelligentagent that takes action on your behalf

o Examples:

Prioritize e-mail Act as gaming partner

Assemble customized news reports

Fill out forms for you

Discuss topics with you

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

MULTI AGENT SYSTEMS AND

8/2/2019 Decision Making and AI

34/38

4-34

MULTI-AGENT SYSTEMS ANDAGENT-BASED MODELING

o Biomimicry learning fromecosystems and adapting theircharacteristics to human andorganizational situations

o Used to

1. Learn how people-based systems behave

2. Predict how they will behave under

certain circumstances3. Improve human systems to make them

more efficient and effective

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

35/38

4-35

Agent-Based Modeling

o Agent-based modeling a way ofsimulating human organizations usingmultiple intelligent agents, each ofwhich follows a set of simple rules andcan adapt to changing conditions

o Multi-agent system groups ofintelligent agents have the ability to

work independently and to interactwith each other

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

36/38

4-36

Business Applications

o Southwest Airlines cargo routingo P&G supply network optimization

o Air Liquide America reduceproduction and distribution costs

o Merck distributing anti-AIDS drugs inAfrica

o Ford balance production costs &

consumer demandso Edison Chouest deploy service and

supply vessels

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

8/2/2019 Decision Making and AI

37/38

4-37

Swarm Intelligence

Swarm (collective) intelligence thecollective behavior of groups of simpleagents that are capable of devisingsolutions to problems as they arise,eventually leading to coherent globalpatterns

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.

Characteristics of Swarm

8/2/2019 Decision Making and AI

38/38

4-38

Characteristics of SwarmIntelligence

o Flexibility adaptable to changeo Robustness tasks are completed

even if some individuals are removed

o Decentralization each individual hasa simple job to do

McGraw-Hill 2007 The McGraw-Hill Companies, Inc. All rights reserved.