Bits and Bytes.docx

Bits and Bytes.docx

《Bits and Bytes.docx》由会员分享，可在线阅读，更多相关《Bits and Bytes.docx（22页珍藏版）》请在冰豆网上搜索。

BitsandBytes

Bits&Bytes:

Lesson4

ComputerIQ

MakingComputersSmart

Doyourecallthisconversationfromthefilm,2001:

ASpaceOdyssey?

Hal,thespacecraftcomputer,killedacrewmemberbythrowinghimintospacewhenthecrewmanwentoutsidetoreplacethecomputerunit.Dave,theheroofthefilm,hasjustdiscoveredthatHallhasalsokilledtherestofthecrewastheyslept.

（Dave）"Openthepodbaydoors,Hal. HelloHal,doyoureadme?

 Doyoureadme,Hal?

 HelloHal,doyoureadme?

Openthepodbaydoors,Hal".

（Hal）"I’msorry,Dave.Ican'tdothat,Dave...themissionistooimportantformetoallowyoutojeopardizeit".

（Dave）"OkHal, I’llgointhroughtheemergencyairlock".

（Hal）"Justwhatdoyouthinkyourdoing,Dave?

 Iknoweverythinghasn’tbeenquiterightwithme,butIcanassureyouit’sgoingtobealrightagain.Ifeelmuchbetternow,Ireallydo."

（Hal）"LookDave,Icanseeyouarereallyupsetaboutthis.IknowI’vemadesomeverypoordecisionslately. Iwanttohelpyou. Stop,Dave…willyoustop,Dave?

I’mafraid,Dave..."

Thisfilm,plusmanybeforeandafter,fueledourlove-fearrelationshipwithsmartmachines.WhilemachineslikeHal,capableofindependentthoughtandemotions,aresciencefiction,artificialintelligenceiscurrentlycombiningcomputerscienceandengineeringtocreatemachinesthatexhibitintelligentbehavior,learnusefulskills,andadapttonewsituations.Thesemachinesareembeddedwithinmanyeverydayappliancesandtoolsinourhomes,hospitals,factories,businesses,governmentagencies,andresearchfacilities.Wholenewfieldsofscienceandbusinessdomainsaredevelopingasnewapplicationsareinventedforthesmartcomputersbehindtheseindispensiblemachines.Andit’sallbaseduponsomeratherbasicprogrammingconceptswhichyouwillcometorealizeinthislesson,areveryhuman.

Expertsystem

-Computerprogramsthatanswerquestionsanddrawconclusionsbyapplyingstatisticalrulesandlogic.

AIandExpertSystems

ConventionalArtificialIntelligence（AI）isthescienceofcreatingcomputersthatmimicwhathumansrefertoasintelligence—theabilitytoprocessinformationandrespondappropriately.AIissupportedbyseveraltypesofmachinelearningtechnologiesincludingexpertsystemswhichapplystatisticalandreasoningcapabilitiestohugeamountsofdatainordertodrawconclusionsandsolveproblems.

Thegoalofexpertsystemsistobringtheknowledgeofexpertstothehandsofdecisionmakerswhoneedaccurateanswersquickly.Computerexpertsystemsaresometimescalledknowledgebasedsystemsandhavebeeninusesincethe1960’s.Theyarecreatedbyfirstgatheringtheknowledgeofhumanexpertsonsomespecific,narrowfieldofknowledge.Then,aknowledgeengineerorganizestheinformationwithrulesthathelptofocustheknowledgetowardsolvingspecificproblems.

Thechildhoodguessinggameinwhichonechildthinksofananimal,andtheotherchildtriestoguessitbyaskingonlyyes-or-noquestions,simulatesanexpertsystem.Aprocessofeliminationbaseduponrulesguidesaseriesofquestionsuntilthecorrectanimalnameisguessed.Itmightgosomethinglikethis:

“Doesithavefur?

”

“No”

“Doesithavefeathers?

”

“Yes”

“Canitfly?

”

“No”

“Isitblackandwhite?

”

“Yes”

“Isitapenguin?

”

“Yes”

FuzzyLogic

-Asystemthatreliesupon“mostlytrue”and“mostlyfalse”conditionstoarriveatmorerealistic“mostlikely”conclusions.

Expertsystemscontaininformationsuppliedbyexpertsonasingletopicandusuallyinvolvea“dialog”withtheuser.Thisproblem-solvingsystemreliesuponrulesthatguidethedialogtowardthediscoveryofthesolutionthatfitsalloftheresponsesgivenbytheuser.Arecentadditiontoexpertsystemsistheinclusionof“fuzzylogic”whichdoesn’trelyondefiniteyes-or-noanswers,butratheronconditionsof“mostlytrue”or“mostlyfalse.”Theselessdefiniteanswersresultinmorerealistic“mostlikely”conclusions.

Expertsystemsareemployedquitecommonlyin:

∙Wizardsthathelpyouinstallsoftware

∙Artificialintelligenceofcomputergames

∙Diagnosticprogramsforeverythingfromidentifyingskindiseasestogettinghelpwithyourcomputersystem

∙Decisionmakinginbusiness,science,engineering,andthemilitary

ComputationalIntelligenceandArtificialNeuralNetworks

Neuralnetwork

-Acollectionofinterconnectedprocessingunitswhichresemblebrainneuronsandarecapableofcomplexanalysisofdata.

Aspecializedareaofartificialintelligencecalledcomputationalintelligenceincludestheapplicationofartificialneuralnetworks.Whengroupsofprocessingunits—sometimesreferredtoasnodes—areinterconnectedinafashionresemblingtheneuronsinthebrain,anartificialneuralnetworkcanbeconstructed.Anartificialneuralnetwork“learns”byobservingcomplexrelationshipsbetweeninputsandoutputsandinferringfromobservations.Thisisparticularlyusefulinapplicationswherethecomplexityofthedataorthedifficultyofthetaskmakestraditionalprogramdesignextremelycomplicatedorimpractical.Artificialneuralnetworksareespeciallyusefulintaskssuchas:

∙

Patternrecognitioninmilitaryradarsystems

∙FaceidentificationinHomelandSecurityoperations

∙Sequencerecognitioningesture,speech,writing

∙MedicaldiagnosiswithEMRtechnology

∙Detectionofabnormalchargesoncreditcards

∙Dataminingorknowledgediscoveryindatabases

∙Identificationofe-mailspam

RoboticsandArtificialIntelligence

Today’srobotsaresimplymachineswithenoughAItodousefultasks.AndthetasksrunthegamutfromentertainmentintheformofAibo,andtheroboticpet,toRoomba,theroboticvacuumcleaner,toPenelope,thesurgicalassistant.Robotscomeinallshapesandsizes.Dependinguponhowyoudefinetheword“robot,”youcouldbetalkingaboutthe“smartwiper”onyourcarwindshieldthatis“justaroboticarm”oramachinethesizeofavehiclesuchastheMarsRoverorthetrucksintheDARPAChallengeroadrace.

Itappearsthatrobotswilldefinitelybepartofourfutureofsmartmachines.Analystsestimatethatabout4millionRoombavacuumrobotswillbeinhomesby2007andsomepeopleinJapanforesee39millionhouseholdrobotsinusebytheendofthedecade.Addtothatthe1.1millionindustrialrobotsintheU.S.andyouhavearobotpopulationgreaterthanthepopulationofCaliforniaorSpain!

ResearchanddevelopmentinAIandrobotsincludessuchprojectsas:

∙

Systemsforsurgicalassistance

∙Solar-poweredunderwatervehicles

∙Roboticassistantsfortheelderly

∙Spaceexplorationrovers

∙Bomb-sniffingmilitaryrobotics

∙Childcareworkers

IDFAmericanAndrosExplosiveOrdinanceDevice（EOD）robot

SincetheearliestdaysofcomputersandAI,greatdebatehasragedoveradefinitionofintelligenceandcreativity,whetherornotamachinecanactuallyunderstandandbeintelligent,andeven,ifthedistinctionbetweenhumanandmachinewilleventuallydisappear.Variousschemestodistinguishbetweencomputerandhumanintelligenceandcreativityhaveappearedoverthepastseveraldecades.PerhapsyouhaveheardoftheTuringTestinwhichajudgeasksquestionstoanentityinanotherroom.Ifatsomepointthejudgecannottelliftheanswersarecomingfromapersonoracomputer,thecomputerisdeclaredintelligent.Andofcourse,we’veallheardofthefamouscomputerversushumanchessgames.Whilephilosopherswillcontinuetoexplorethedistinctionsbetweenhumansandmachines,computerscientistswillforgeaheadwithresearchanddevelopmentsofmoreandmoreamazingmachines.Ultimately,computerIQcomesdowntomachinesfollowingthedirectionsgiventoitbyprogrammersandengineers.

GetReal

1.ExploreanexpertsystembyvisitingTheWhaleWatcherathttp:

//www.aiinc.ca/demos/whale.html.Doesthissystememploystricttrue/falselogicorfuzzylogic?

2.Reviewthebasiccharacteristicsofexpertsystems,fuzzylogic,neuralnetworks,androbots.Matchthetasksontheleftwiththetechnologyontheleftthatwouldbestbeabletoaccomplishthetaskdescribed.

Tasktoaccomplish

Technology

____Discoverthemostlikelycauseofapanicattack

A.Expertsystems

____Identifywildflowersseenonahike

B.Fuzzylogic

____Inspectminesfordangerousgases

C.Neuralnetworks

____Uncoverforgedartwork

D.Robotics

Herearesomepossibleanswers:

1.Afteridentifyingseveralhypotheticalsightingofwhales,itappearsthatthissystememploysstricttrue/falselogic.Alloftheresponsechoicesareyes/nooreither/or.

2.

Tasktoaccomplish

Technology

__B_Discoverthemostlikelycauseofapanicattack

A.Expertsystems

A_Identifywildflowersseenonahike

B.Fuzzylogic

__D_Inspectminesfordangerousgases

C.Neuralnetworks

__C_Uncoverforgedartwork

D.Robotics

TeachingComputerstoMakeDecisions

Computersare“taught”todisplayintelligencebymakingdecisionsthroughvariouscodingtechniques.Collectivelythesetechniquesarecalled“selectionstructures”;thelinesofcodeinstructthecomputerto“select”betweenvariousoptionsbaseduponparticularconditions.

If-Statements

If-statement

-Acomputerinstructionthatdecidesuponanactionbasedupontheevaluationofacondition.

Themostbasicselectionstructureistheif-statement.Inthereal-worldweprocessif-statementssooftenandwithsuchspeedthatwebarelygiveitanythoughtatall.Youhavelikelyexperiencedtheseopportunitiestoprocessan“if”decision:

Ifit’srainingoutside,yougrabanumbrella.

Ifsteakisonsale,youbuyit.

Ifthetrafficsignalisred,youstop.

Ifyouseeafriend,youshout,“Hi!

”

Inallofthese