Slide 1AdaptationClasses of AdaptationNo AdaptationSubstitution AdaptationExampleExample (2)Substitutional Feedback-basedSubstitutional Constraint-basedCompositional AdaptationAdaptation Operators (2)Case AdaptationSources:–Chapter 8–www.iiia.csic.es/People/enric/AICom.html–www.ai-cbr.orgAdaptationAdaptationNew problemSelected caseAdaptation knowledgeSolutionClasses of Adaptation•No-adaptation•Transformational AnalogySubstitution AdaptationFeedback basedconstraint basedCompositional adaptation•Generative Solution AdaptationTransformational AnalogyDerivational AnalogyNo Adaptation •For classification/diagnosis tasks•If C is the NN for a new problem P then class(P) class(C)•If {C1,…,Ck} are the k-NN for a new problem P thenClass(C) F({C1,…,Ck})For example: F({C1,…,Ck}) = “majority class among {C1,…,Ck}”Substitution Adaptation•Let C = (P,S); A problem P and a solution S•Adaptation problem:GivenA problem P’A case C such that P is similar to P’Search a substitution such that (S) solves P’ corresponds to an application of a rule transforming parts of the case (so it is not a substitution in the traditional sense)ExampleSupport for PC sale:•Cases are configuration episodes of PCs•User specifies his/hers requirements•System selects best PC (e.g., using CCBR) and change some componentsExample rules (Substitutional Adaptation): If (query.application = ‘database’ and case.diskSpace < 2GB) then target.diskSpace 4GBExample (2)Example rules (Substitutional Adaptation): If (query.application = ‘games’ and case.application ‘games’) then AddObject target.addJoystick AddObject target.addSoundCardOther rules to configure joystick and soundSubstitutional Feedback-basedCar type: sportColor: redSeating: 2Valves: 48Type: 5.7LModel name: name1Price: 200,000 Year: 2003 Feedback: not successfulCause: price is too highCar type: sportColor: redSeating: 2Valves: 48Type: 5.7LModel name: name1Price: 200,000 Year: 2003Feedback: successfulCar type: sportColor: redSeating: 2Valves: 40Type: 3.6LModel name: name 2Price: 150,000 Year: 2000Feedback: successfulAdaptCaseC (adapted)CaseA (new)CaseB (old)RetrieveCopyRule:if price is too highthen model previous modelSubstitutional Constraint-basedCase ID: 123Speed: highPrice: middleUsage: sportAntitheft performance: highModel Name: Toyota Sedan 07Price: 10,500Antitheft system: Product ACase ID: 456Speed: highPrice: middleUsage: sportAntitheft performance: middleModel Name: Toyota Sedan 07Price: 10,500Antitheft system: Product ACase ID: 123Speed: highPrice: middleUsage: sportAntitheft performance: high Model Name: Toyota Sedan 07+Price: 11,000Antitheft system: Product BCaseA (new)CaseB (old)RetrieveCopyadaptCaseC (adapted)Rule:if need higher Antitheft performance and Antitheft System = product Athen Antitheft System product B Price Price + 500Compositional Adaptation•Let C = (P,S); A problem P and a solution S•Adaptation problem:GivenA problem P’A case C such that P is similar to P’Search a sequence of substitutions 1, …, n such that:S’ is a solution for P’(P,C)… (P’,S’)12nAdaptation Operators (2)Uses rule-based systems during adaptationRoles of operators/rules:• General knowledge about the domain• (P,C)… (P’,S’)12nAdaptation
View Full Document