您好,欢迎访问三七文档
TitleofPublication:EncyclopediaofCognitiveScience.Articletitle:ConstraintSatisfaction.Articlecode:26.Authors:RinaDechterDepartmentofComputerandInformationScienceUniversityofCalifornia,IrvineIrvine,California,USA92717Telephone:949-824-6556Fax:949-824-4056E-mail:dechter@ics.uci.eduFrancescaRossiDipartimentodiMatematicaPuraedApplicataUniversit adiPadovaViaBelzoni7,35131Padova,ItalyTelephone:+39-049-8275982Fax:+39-049-8275892E-mail:frossi@math.unipd.it1ConstraintSatisfactionKeywords:Constraints,constraintsatisfaction,constraintpropagation,con-straintprogramming.Contents1Introduction32Constraintpropagation53Constraintsatisfactionassearch74Tractableclasses95Constraintoptimizationandsoftconstraints106Constraintprogramming117Summary13Articlede nition:Constraintsareadeclarativeknowledgerepresentationformalismthatallowsforacompactandexpressivemodelingofmanyreal-lifeproblems.Constraintsatisfactionandpropagationtools,aswellasconstraintprogramminglanguages,aresuccessfullyusedtomodel,solve,andreasonaboutmanyclassesofproblems,suchasdesign,diagnosis,scheduling,spatio-temporalreasoning,resourceallocation,con guration,networkoptimization,andgraphicalinterfaces.21IntroductionConstraintsatisfactionproblems.Aconstraintsatisfactionproblem(CSP)consistsofa nitesetofvariables,eachassociatedwithadomainofvalues,andasetofconstraints.Eachoftheconstraintisarelation,de nedonsomesubsetofthevariables,calleditsscope,denotingtheirlegalcombinationsofvalues.Aswell,constraintscanbedescribedbymathematicalexpressionsorbycomputableprocedures.Solutions.Asolutionisanassignmentofavaluetoeachvariablefromitsdomainsuchthatalltheconstraintsaresatis ed.Typicalconstraintsatisfactionproblemsaretodeterminewhetherasolutionexists,to ndoneorallsolutions,andto ndanoptimalsolutionrelativetoagivencostfunction.ExamplesofCSPs.Anexampleofaconstraintsatisfactionproblemisthewell-knownk-colorabilityproblem.Thetaskistocolor,ifpossible,agivengraphwithkcolorsonly,suchthatanytwoadjacentnodeshavedi erentcolors.Aconstraintsatisfactionformulationofthisproblemassociatesthenodesofthegraphwithvariables,thepossiblecolorsaretheirdomainsandthenot-equalconstraintsbetweenadjacentnodesaretheconstraintsoftheproblem.Anotherknownconstraintsatisfactionproblemconcernssatis ability(SAT),whichisthetaskof ndingatruthassignmenttopropositionalvariablessuchthatagivensetofclausesaresatis ed.Forexample,giventhetwoclauses(A_B_:C);(:A_D),theassignmentoffalsetoA,truetoB,falsetoC,andfalsetoDisasatisfyingtruthvalueassignment.Theconstraintgraph.Thestructureofaconstraintproblemisusuallydepictedbyaconstraintgraphwhosenodesrepresentsthevariables,andanytwonodesareconnectedifthecorrespondingvariablesparticipateinthesameconstraintscope.Inthek-colorabilityformulation,thegraphtobecoloredistheconstraintgraph.IntheSATexampleabove,theconstraintgraphhasAconnectedwithD,andA;BandCareconnectedtoeachother.3Applicationsareas.Constraintproblemshaveprovensuccessfulinmod-elingmundanecognitivetaskssuchasvision,languagecomprehension,de-faultreasoningandabduction,aswellasinapplicationssuchasscheduling,design,diagnosis,andtemporalandspatialreasoning.Thereasonisthatconstraintsallowforanatural,expressiveanddeclarativeformulationofwhathastobesatis ed,withouttheneedtosayhowithastobesatis ed.Complexityofconstraint-relatedtasks.Ingeneral,constraintsatis-factiontasks(like ndingoneorallsolutions,orthebestsolution)arecom-putationallyintractable(NP-hard).Intuitively,thismean,thatintheworstcaseallthepossiblevariableinstantiationsmayneedtobeconsideredbe-foreasolution(orbestsolution)canbefound.However,therearesometractableclassesofproblemsthatallowfore cientsolutionalgorithmsevenintheworst-case.Moreover,alsofornon-tractableclasses,manytechniquesexhibitagoodperformanceinpracticeintheaveragecase.Constraintoptimizationproblems.Constraintprocessingtasksincludenotonlythesatisfactiontask,butalsoconstraintoptimizationproblems.Thisoccurswhenthesolutionsarenotequallypreferred.Thepreferencesamongsolutionscanbeexpressedviaanadditionalcostfunction(alsocalledanobjectivefunction),andthetaskisto ndabest-costsolutionorarea-sonableapproximationofit.Softconstraints.Thenotionofconstraintoptimizationleadsamore ex-ibleviewofconstraints,whereeachconstraintmayhavealevelofimpor-tance.Inthiscase,wetalkaboutsoftconstraints,whichallowforafaithfulmodelingofmanyapplicationsandcanaccommodateuserpreferencesanduncertaintieswithintheconstraintformalism.TechniquesforsolvingCSPs.Thetechniquesforprocessingconstraintproblemscanberoughlyclassi edintotwomaincategories:searchandcon-sistencyinference(orpropagation).However,suchtechniquescanalsobecombined,and,infact,inpracticeaconstraintprocessingtechniqueusuallycontainsaspectsofbothcategories.Searchalgorithmstraversethespaceofpartialinstantiations,buildingupacompleteinstantiationthatsatis esalltheconstraints,ortheydeterminethattheproblemisinconsistent.In4contrast,consistency-inferencealgorithmsreasonthroughequivalentprob-lems:ateachsteptheymodifythecurrentproblemtomakeitmoreexplicitwithoutloosinganyinformation(thatis,maintainingth
本文标题:Title of Publication Encyclopedia of Cognitive Sci
链接地址:https://www.777doc.com/doc-3093837 .html