《管理信息系统 MIS课件：ch09 Complex Decisions and Artificial Intelligence》由会员分享，可在线阅读，更多相关《管理信息系统 MIS课件：ch09 Complex Decisions and Artificial Intelligence（48页珍藏版）》请在金锄头文库上搜索。
1、Introduction to MIS1Copyright 1998-2002 by Jerry PostIntroduction to MISChapter 9Complex Decisions and Artificial IntelligenceIntroduction to MIS3Computer analysisof data and model.DecisionOperationsTacticsStrategyNeural networkCompanyComplex Decisions& Artificial IntelligenceIntroduction to MIS4Out
2、lineSpecialized ProblemsExpert SystemvDefinitionvBuilding Expert SystemsvDSS and ESKnowledge ManagementAIMore Specialized ProblemsDSS, ES, and AIE-Business and Software AgentsConclusionCases: FranchisesAppendix: E-mail RulesIntroduction to MIS5Requirement for Decision Making:Specialized ProblemsDiag
3、nosticsSpeedConsistencyTrainingCase-based reasoningIntroduction to MIS6Expert SystemKnowledge BaseSymbolic & Numeric KnowledgeIf income 20,000or expenses 3000and good credit historyor . . .Then 10% chance of defaultRulesExpert decisionsmade bynon-expertsExpertHow to manage rules?The decision tree?In
4、troduction to MIS7Expert System ExampleCamcorder selection by ExSysLink: http:/ ItMore: Pallas food:Map: http:/ accordAirplaneIntroduction to MIS8ES Example: bank loanWelcome to the Loan Evaluation System.What is the purpose of the loan? carHow much money will be loaned? 10,000For how many years? 5T
5、he current interest rate is 10%.The payment will be $212.47 per month.What is the annual income? 24,000What is the total monthly payments of other loans? Why?Because the payment is more than 10% of the monthly income.What is the total monthly payments of other loans? 50.00The loan should be approved
6、, there is only a 2% chance of default.Forward ChainingIntroduction to MIS9Payments 10%monthly income?Other loanstotal k cluster-size) do (for (j 0 (+ 1 j ) exit when (= j k) do (connect unit cluster k output o -A to unit cluster j input i - A ) . . . )Maintained by expert system shellProgrammerCust
7、om program in LISPES DevelopmentES ShellsvGuruvExsysCustom ProgrammingvLISPvPROLOGIntroduction to MIS15DSS and ESIntroduction to MIS16ES Problem SuitabilityNarrow, well-defined domainSolutions require an expertComplex logical processingHandle missing, ill-structured dataNeed a cooperative expertRepe
8、atable decisionIntroduction to MIS17Limitations of ESFragile systemsvSmall environmental. changes can force revision. of all of the rules.MistakesvWho is responsible?Expert?Multiple experts?Knowledge engineer?Company that uses it?Vague rulesvRules can be hard to define.Conflicting expertsvWith multi
9、ple opinions, who is right?vCan diverse methods be combined?Unforeseen eventsvEvents outside of domain can lead to nonsense decisions.vHuman experts adapt.vWill human novice recognize a nonsense result?Introduction to MIS18Knowledge ManagementA collection of a documents and datavCreated by expertsvS
10、earchablevWith links to related topicsvHighly organized groupwareEmphasizing contextMain problemconvincing everyone to enter and update the documentsExampleNew: ontology and semanticsIntroduction to MIS20AI Research AreasComputer SciencevParallel ProcessingvSymbolic ProcessingvNeural NetworksRobotic
11、s ApplicationsvVisual PerceptionvTactilityvDexterityvLocomotion & NavigationNatural LanguagevSpeech RecognitionvLanguage TranslationvLanguage ComprehensionCognitive SciencevExpert SystemsvLearning SystemsvKnowledge-Based SystemsIntroduction to MIS21More Specilised ProblemsPattern RecognitionMachine
12、VisionVoice and Speech RecognitionLanguage ComprehensionRobotics and MotionStatistics, Uncertainty, and Fuzzy LogicIntroduction to MIS22Output CellsSensory Input CellsHidden LayerSome of the connections3-274Input weightsIncompletepattern/missing inputs.Neural Network: Pattern recognitionIntroduction
13、 to MIS23Machine Vision ExampleThe Department of Defense has funded Carnegie Mellon University to develop software that is used to automatically drive vehicles. One system (Ranger) is used in an army ambulance that can drive itself over rough terrain for up to 16 km. ALVINN is a separate road-follow
14、ing system that has driven vehicles at speeds over 110 kph for as far as 140 km.DASO的项目三荾的项目Introduction to MIS24Speech RecognitionLook at the users voice command:Copy the red, file the blue, delete the yellow mark.Now, change the commas slightly.Copy the red file, the blue delete, the yellow mark.I
15、 saw the Grand Canyon flying to New York.EmergencyVehiclesNoParkingAny TimeIntroduction to MIS25Subjective Definitionstemperaturereference pointe.g., averagetemperaturecoldhotMoving farther from the reference pointincreases the chance that the temperature isconsidered to be different (cold or hot).S
16、ubjective (fuzzy) DefinitionsIntroduction to MIS26DSS, ES, and AI: Bank ExampleDecision Support SystemExpert SystemArtificial IntelligenceName Loan#Late AmountBrown 25,000 51,250Jones 62,000 1 135Smith 83,000 32,435.DataIncomeExisting loansCredit reportModelLend in all but worst casesMonitor for lat
17、e and missing payments.OutputES RulesWhat is the monthly income?3,000What are the total monthly payments on other loans? 450How long have they had the current job? 5 years. . .Should grant the loan since there is only a 5% chance of default.Determine Rulesloan 1 data: paidloan 2 data: 5 lateloan 3 d
18、ata: lostloan 4 data: 1 lateData/Training CasesNeural Network WeightsEvaluate new data,make recommendation.Loan OfficerIntroduction to MIS28Vacation ResortsSoftware agentResort DatabasesLocate &book trip.Software AgentsO-O based, an objectIndependentNetworks/CommunicationUsesvSearchvNegotiatevMonito
19、rSOAIntroduction to MIS29Conclusion and AI QuestionsDSS, ES, KM, AI, the definition and features High level of Computers ApplicationsWhat is intelligence?vCreativity?vLearning?vMemory?vAbility to handle unexpected events?vMore?Can machines ever think like humans?How do humans think?Do we really want
20、 them to think like us?Introduction to MIS31Cases: FranchisesIntroduction to MIS32Cases: Mrs. FieldsBlockbuster VideoWhat is the companys current status?What is the Internet strategy?How does the company use information technology?What are the prospects for the industry?Introduction to MIS33Appendix
21、: E-Mail Rules - FoldersFolders make it easy to organize and handle your mail.Simple rules from the Tools + Organize button move messages directly to the specified folder.Introduction to MIS34Rules: ConditionsThe Tools + Rules Wizard makes it easy to create rules. Begin with a blank rule.Set the Con
22、ditionsSet the ActionsDefine ExceptionsA sample rule to handle unsolicited credit card applications.Introduction to MIS35Rules: ActionsChoose an action.You can choose multiple actions, but be careful. The marking options are often combined.Introduction to MIS36Rules: ExceptionsRules can have excepti
23、ons. For example, you might want to delete company newslettersunless one has your name in it.Introduction to MIS37Rule Sequences: Decision TreeFrom boss, Subject: ExpensesMessage from ExpenseAccountingExpenses FolderSet expenses categoryMove itRule 1Rule 2Expenses categorySubject: PaymentRule 3Actio
24、n: Mark important and notify.决策运作层战术层战略层神经网络公司公司知识库符号及数字信息如果如果 income 20,000或者 expenses 收入的 5% 计算其它贷款支付的总额 计算月收入的支付百分率 如果该百分率 25%如果新的贷款 10%，则发放贷款否则 如果月花费总额 收入的 40%，则发放贷款 否则 如果 50%，而且是5年以上的客户， 或者 60%，而且是10年以上的客户，而且在同一居所居住了5年以上，则发放贷款支付是否 月收入的10%?其它贷款总额是否 月收入的 10% 其它贷款的月支付总额是多少？ 50.00贷款可以批准，不被批准的概率仅为贷款
25、可以批准，不被批准的概率仅为 2%用户使用的专家系统的显示器由设计人员输入的规则及决策树专家专家系统外壳的双向推理知识工程师知识数据库(for (k 0 (+ 1 k) ) exit when ( ? k cluster-size) do (for (j 0 (+ 1 j ) exit when (= j k) do (connect unit cluster k output o -A to unit cluster j input i - A ) . . . )由专家系统外壳进行维护程序员LISP 的客户程序输出层细胞输出层细胞感觉输入层细胞感觉输入层细胞隐藏层隐藏层一些联接3-274输入
26、权重不完整模式 /丢失输入语言理解实例语言理解实例看看当你给计算机第一组指令时会怎么样？假设它没有正确地听到逗号，从而产生了第二句的理解：1.copy the red， file the blue， delete the yellow mark.2.copy the red file， the blue delete， the yellow mark.看看下面的句子，人是可以理解的，但要计算机来就文字做翻译，会是什么意思呢？“I saw the Grand Canyon flying to New York”人的理解：我在飞往纽约的途中看到了大峡谷计算机的理解：我看见了飞越纽约的大峡 温度基准
27、点例如：平均温度冷热从基准点开始移动，温度的变化会增加，会看到温度的不同（冷或热）。决策支持系统决策支持系统专家系统专家系统人工智能人工智能Name Loan#Late AmountBrown 25,000 51,250Jones 62,000 1 135Smith 83,000 32,435.数据收入已有贷款信用记录模式贷款给所有的人是最坏的情况监视晚支付及丢失支付的情况输出专家系统规则专家系统规则月收入是多少？ 3,000其它贷款的月支付总额是多少？ 450现有的工作持续了多久？ 5 years. . .由于不被批准的概率仅为 5%，故贷款应该被批准确定规则确定规则loan 1 data: paidloan 2 data: 5 lateloan 3 data: lostloan 4 data: 1 late数据 / 训练案例神经网络权重评估新数据，得出推论.信贷员信贷员度假胜地软件代理度假胜地数据库地点及旅行预定
