《人工智能一种现代方法》第四版习题答案docx.docx
- 文档编号:6438458
- 上传时间:2023-01-06
- 格式:DOCX
- 页数:16
- 大小:35.45KB
《人工智能一种现代方法》第四版习题答案docx.docx
《《人工智能一种现代方法》第四版习题答案docx.docx》由会员分享,可在线阅读,更多相关《《人工智能一种现代方法》第四版习题答案docx.docx(16页珍藏版)》请在冰豆网上搜索。
《人工智能一种现代方法》第四版习题答案docx
Chapter1
LIDefineinyourownword:
(a)intelligence,(b)artificialintelligence,(c)agent.
•IntelligenceDictionarydefiiutionsofmtelligencetalkabout"thecapacitytoacquueandapplyknowledge^or"thefacultyofthoughtandreason^or"theabilitytocomprehendandprofitfromexperience.^Theseareallreasonableanswers,butifwewantsomethingquantifiablewewouldusesomethinglike"theabilitytoapplyknowledgeinordertoperformbetterinanenvironment
智能的孝典定义有一种学习或应用知识M能力,一种思考和推理的本领,领会并且得益于经验的能力,这些都是有道理的答案,但如果我们想量化一些东西,我们将用到一些东西像为了在环境中更好的完成任务使能力适应知识
•Artificialintelligence人工智能:
Wedefineartificialintelligenceasthestudyandconstructionofagentprogramsthatperfoimwellinagivenenviiomnent.foragivenagentarchitectiue・
作£一学习和构造智能体程序,为了一个智能体结构,在被给的环境中可以很好的完成任务。
•Agen智能体1:
Wedefineanagentasanentity实彳本thattakesactioninresponsetoperceptsfromanenvironment.在一个环境中对一个对象做出反应的实体
1.4Therearewell-knownclassesofproblemthatareintractablydifficultforcomputers,andotherclassesthatareprovablyundecidable.DoesthismeanthatAlisimpossible?
No.ItmeansthatAlsystemsshouldavoidtrymgtosolvemtractableproblems・Usually,thismeanstheycanonlyappioxiniateoptmialbeliavioi.Noticetliatliuniansdon'tsolveNPcompletepiobleniseither.Sometimestheyaregoodatsolvingspecificmstanceswithalotofstructure,perhapswiththeaidofbackgroundknowledge.Alsystemsshouldattempttodothesame・
Lil“surelycomputerscannotbeintelligent-theycandoonlywhattheirprogrammerstellthem.”Isthelatterstatementtrue,anddoesitimplytheformer?
Thisdependsonyourdefinitionof"intelligent^and"tell."Inonesensecomputersonlydowhatthepiogramnierscommandthemtodo,butinanothersensewliattheprograniniersconsciouslytellsthecomputertodooftenliasverylittletodowithwhatthecomputeractuallydoes・Anyonewhohaswrittenaprogramwithanornerybugknowsthis,asdoesanyonewhohaswrittenasuccessfulmachinelearningprogram・SoinonesenseSamuel"told、'thecomputer"learntoplaycheckersbetterthanIdo.andthenplaythatway/'butmanothersensehetoldthecomputer“followthislearnuigalgontlim'anditlearnedtopby.Sowereleftinthesituationwhereyoumayormaynotconsiderleanungtoplaycheckerstobessigilofintelligence(oryoumaythinkthatlearningtoplayintherightwayrequiresintelligence,butnotmthisway),andyoumaythinkthemtelligenceresidesintheprogrammerorinthecomputer
Chapter2
2・1Defineinyourownwordsthefollowingterms:
agent,agentfunction,agentprogram,rationality,reflexagent,model-basedagent,goal-basedagent,utility-basedagent,learningagent.
Thefollowmgaiejustsomeoftheniaiivpossibledefinitionstliatcanbevvntten:
•Agent智能体:
anentity(实体)tliatperceives(感知)andactsIf为;ot,onethatcanbeviewedasperceivingandactmg・Essent丄ally木质上anyobjectqualifies,^>12;thekeypointistliewaytheobjectimplementsanagentfbnchon.(Note:
someauthorsrestrictthetermtoprogramsthatoperateonbehalfof2human,ortoprogramsthatcancausesomeoralloftheircodetorunonothermachinesonanetwork,asinmobileagents.MOBILEAGENT)
一个具冇感知和行文的实体,或者是一个可以观察到感觉的实体,本质上,任何限定对彖,只要的观点是一种对象执行智能体函数的方法。
(注意,一些作者)
可以感知环境,并在环境中行动的某种东西。
•AgentJimction智能体函数:
afunctionthatspecifiestheagent'sactioninresponsetoeverypossibleperceptsequence.智能体相应任何感知序列所采収的彳亍动
•Agentprogram智能体thatprogramwhich,combinedwithamachinearchitecture,implementsanagentfunction.Inoursimpledesigns,theprogramtakesanewperceptoneachinvocationandreturnsanaction.实现了智能函数。
有各种基本的智能体程序设计,反应出现实衣现的•级用丁•决策过程的信息种类。
设计可能在效率、压缩性和灵活性方而冇变化。
适出的智能体程序设计取决于环境的本性
•Rationality;理性:
apropertyofagentsthatchooseactionsthatmaximizetheirexpectedutility,giventheperceptstodate・
•Autonomy自主:
apropertyofagentswhosebehaviorisdeteimniedbytheirownexperiencelatherthansolelybytheiruutialprogianmung・
•Reflexagenf反射型智能体:
anagentwhoseactiondependsonlyonthecurrentpercept.
一个智能体的行为仅仅依赖于当前的知觉。
•Model-basedagent基于模型的智能体^ailagentwhoseactionisderivedduectlyfromaninternalmodelofthecurrentworldstatethatisupdatedovertune.
一个解能体的行为育接得自于内在模世的状态,这个状态是半前世界通用的不断更新。
•Goal-basedage"基于目标的智能体t:
anagenttliatselectsactionsthatitbelieveswillachieveexplicitlyrepresentedgoals.®能体选择它相信能明确达到目标的行动。
•Utility-basedagen基于效用的智能体t:
anagentthatselectsactionsthatitbelieveswillmaximizetheexpectedutilityoftheoutcomestate.试图最大化他们白己期卑的快乐
•Learningagenl学习智能体:
anagentwhosebehaviormiprovesovertunebasedonitsexperience・
2.2Boththeperformancemeasureandtheutilityfunctionmeasurehowwellanagentisdoing.Explainthedifferencebetweenthetwo.
Aperfoiniaiicemeasure(ftfit/x)isusedbyanoutsideobserve】toevaluate(14竹占)howsuccessfulanagentis.Itisafunctionfromlustonestoarealnumber・Autilityftinction(效用函数)isusedbyanagentitselftoevaluatehowdesirable(令人想要)statesorhistonesare.I11ourfiamework,theutilityfunctionmaynotbethesameastheperformancemeasure;furthermore,anagentmayhavenoexplicitutilityfunctionatall,whereasthereisalwaysaperfoiniancemeasure・
2.5Foreachoffollowingagents,developaPEASdescriptionofthetaskenvironment:
a.Robotsoccerplayer;
b>Internetbook-shoppingagent;
c.AutonomousMarsrover;
d.Mathematicianstheorem-provingassistant.
Someirpresentative.butnotexhaustive,answersaregiveninFigiueS2.1.
AgentTyi)e
PeifonnanceMeasiuc
Enviroiunent
Acniators
Sensors
Robotsoccerplayer
Wiiuiiiiucauie.
C7J*
goalsfex/against
Field,ball.oi\nteam,otherteamownbody
Devices(e.g・,lp>)forlocomotionandkicking
Camera,touchsensors,accclcroincrci's.oiicntationsensors,wheeljointencoders
Internetbook-shoppingagent
Obtainrequestedinterestingbooks・minimizeexpenditure
Inreniet
Followlink,entei'submitdatainfields,displaytouser
Webpages,userrequests
AutonomousMarsrover
Terrainexploredandreportedsampleseatheredandanalyzed
Launchvehicle,lander.Mars
Wheel、legs,samplecollectiondevice,analysisdevices,radiotransmitter
CanieiTi.touchsensors,acceleromereis.oiientationsensors,.whccl/jointencoders,radioreceiver
Mathematician\thcoicnvprovingassistant
FigureS2.1AgenttypesandtlieirPEASdescriptions,forEx.2・5・
智能体类型
性能度最
坏境
执行器
传感器
机器人足
贏得比赛,
裁判,自己队伍,
相机,触摸传感器
球运动员
打败对手
英他队伍,门己身体
装宜(腿)行走踢球
加速器
1大1特网购
获得请求/感兴趣
向卜连接,输入提
书智能体
的书,绘小支出
因特网
交数据,用户显示器
网页,用户请求
自主火星
地形探测,汇报,
火星,运行装置,
轮子/腿,简单手机装置,
相机,触摸传感器
漫步者
样本采集分析
登陆器
分析装登,无线电发射装盘
方向传感器
数学家的定理
证明助手
2.6ForeachoftheagenttypeslistedinExercise2・5,characterizetheenvironmentaccordingtothepropertiesgiveninSection2・3andselectasuitableagentdesign.Thefollowingexercisesallconcerntheimplementationofenvironmentandagentsforthevacuum-cleanerworld.
EnvironmentpropertiesaiegiveninFigureS2.2・Suitableagenttypes:
a.Amodel-basedreflexagentwouldsufficeformostaspects;fortacticalplay,autilitybasedagentwithlookaheadwouldbeuseful.^于模里的映射能够满足大多数要求,对于战术游戏,向前效川智能体将会有用
b.Agoal-basedagentwouldbeappiopnateforspecificbookrequests・Formoreopenendedtasks―.g.,"Findmesomethmgmterestmgtoread"—tradeoffs(权衡折中)meinvolved(棘于•的)andtheagentmustcompareutilitiesforvarious(不同的)possiblepurchases.^于H标的fV能体将适、"l的明确书的in求,为史幺开放的任务,例如伶找我有兴趣渎的巧,智能体必须比较冷种可能的购买方式之间的效川
c.Amodel-basedreflexagentwouldsufficeforlow-levelnavigationandobstacleavoidance;forrouteplanning,explorationplanning,expeimientation,etc.,somecombinationofgoal-basedandutility-basedagentswouldbeneeded.基于模型的映射智能休能够满足低水平的航线和避免障碍,为了路由汁划,ft:
测计划,实验等。
这需要基于目标和效用的智能体。
d・Forspecificprooftasks,agoal-basedagentisneeded.For"exploratorytasks―・g・、"Trovesomeusefullemmataconcenungoperationsonstrings—autility-basedarchitecturenuglitbeneeded・
为了明确的检验任务,需要基于目标的智能体,为探测任务,
TaskEnvHonnient
Observable
Detenmiustic
Episodic
Static
Discrete
Agents
Robotsoccer
Partially
Stochastic
SequentialDynamic
Continuous
Multi
Internetbook-shoppine
Paitially
Deterministic*
Sequential
Static*
Discrete
Single
AutonomousMarsrover
Paitially
Stochastic
SequentialDynamic
Continuous
Single
Mathematician'sassistant
Fully
Determuustic
Sequential
Semi
Discrete
Multi
FigureS2.2EmiromnenrpropertiesforEx.2.6.
任务环境
可观察性
确定性
片段性
静态性
离散型
智能体数
机器人足
球运动员
部分
随机的
连续的
动态的
连续的
多
因特网购
书智能体
部分
确定的
连续的
静态的
离散的
单
自主火星
漫步者
部分
随机的
连续的
动态的
连续的
单
数学家的定理
证明助手
完全
确泄的
连续的
静态的
离散的
多
Chapter3
3.1Defineinyourownwordsthefollowingterms:
state,statespace,searchtree,searchnode,goal,action,successorfunction,andbranchingfactor.
•state:
Astateisasituationthatanagentcanfinditselfin.Wedistinguishtwotypesofstates:
worldstates(theactualconcretesituationsintherealworld)andrepresentationalstates(theabstractdescriptionsoftherealworldthatareusedbytheagentindeliberatingaboutwhattodo).
•statespace;Astatespaceisagraphwhosenodesarethesetofallstates,andwhoselinksaieactionsthattransformonestateintoanother・
•searchtree:
Asearchtreeisatree(agiaphwithnoundirectedloops)inwhichtherootnodeisthestartstateandthesetofchildrenforeachnodeconsistsofdiestatesreachablebytakinganyaction.
•searchnode;Asearchnodeisanodemthesearchtiee・
•goal;Agoalisastatethattheagentistryingtoreach・
•action;Anactionissomethingthattheagentcanchoosetodo.
•successorfu
- 配套讲稿:
如PPT文件的首页显示word图标,表示该PPT已包含配套word讲稿。双击word图标可打开word文档。
- 特殊限制:
部分文档作品中含有的国旗、国徽等图片,仅作为作品整体效果示例展示,禁止商用。设计者仅对作品中独创性部分享有著作权。
- 关 键 词:
- 人工智能一种现代方法 人工智能 一种 现代 方法 第四 习题 答案 docx
![提示](https://static.bdocx.com/images/bang_tan.gif)