Solution Manual for Artificial Intelligence A Modern Approach, 4th Edition by Stuart Russell and Peter Norvig

SOLUTIONS & INSTRUCTOR MANUAL
l l l




Artificial Intelligence: A Modern Approach, 4th Edition
l l l l l l




by Peter Norvig and Stuart Russell, Chapters 1 – 28
l l l l l l l l l l

,Artificial Intelligencel




1 Introduction ...
llll l l l




2 Intelligent Agents ...
llll l l l l




II Problem-solving
l




3 Solving Problems by Searching ...
llll l l l l l l




4 Search in Complex Environments ...
llll l l l l l l




5 Adversarial Search and Games ...
llll l l l l l l




6 Constraint Satisfaction Problems ...
llll l l l l l




III Knowledge, reasoning, and planning
l l l l




7 Logical Agents ...
llll l l l l




8 First-Order Logic ...
llll l l l l




9 Inference in First-Order Logic ...
llll l l l l l




10 Knowledge Representation ...
llll l l l l




11 Automated Planning ...
llll l l l l




IV Uncertain knowledge and reasoning
l l l l




12 Quantifying Uncertainty ...
llll l l l l




13 Probabilistic Reasoning ...
llll l l l l




14 Probabilistic Reasoning over Time ...
llll l l l l l l




15 Probabilistic Programming ...
llll l l l l




16 Making Simple Decisions ...
llll l l l l l




17 Making Complex Decisions ...
llll l l l l l




18 Multiagent Decision Making ...
llll l l l l l




V Machine Learning
l l

, 19 Learning from Examples ...
llll l l l l l




20 Learning Probabilistic Models ...
llll l l l l l




21 Deep Learning ...
llll l l l l




22 Reinforcement Learning ...
llll l l l l




VI Communicating, perceiving, and acting
l l l l




23 Natural Language Processing ...
llll l l l l l




24 Deep Learning for Natural Language Processing ...
llll l l l l l l l l




25 Computer Vision ...
llll l l l l




26 Robotics ...
llll l l l




VII Conclusions
l




27 Philosophy, Ethics, and Safety of AI ...
llll l l l l l l l l




28 The Future of AI
llll l l l l

, EXERCISES l l

1
INTRODUCTION
Note lthat lfor lmany lof lthe lquestions lin lthis lchapter, lwe lgive lreferences lwhere lanswers lcan lbe
lfound lrather lthan lwriting lthem lout—the lfull lanswers lwould lbe lfar ltoo llong.




1.1 What Is AI?
l l l




Exercise l1.1.#DEFA
Define lin lyour lown lwords: l (a) lintelligence, l(b) lartificial lintelligence, l(c) lagent, l(d) lra-
ltionality, l(e) llogical lreasoning.




a. Dictionary ldefinitions lof lintelligence ltalk labout l“the lcapacity lto lacquire land lapply
lknowledge” lor l“the lfaculty lof lthought land lreason” lor l“the lability lto lcomprehend land

lprofit lfrom lexperience.” l These lare lall lreasonable lanswers, lbut lif lwe lwant lsomething

lquantifiable lwe lwould luse lsomething llike l“the lability lto lact lsuccessfully lacross la lwide

lrange lof lobjectives lin lcomplex lenvironments.”

b. We ldefine lartificial lintelligence las lthe lstudy land lconstruction lof lagent lprograms lthat
lperform lwell lin la lgiven lclass lof lenvironments, lfor la lgiven lagent larchitecture; lthey ldo

lthe lright lthing. l An limportant lpart lof lthat lis ldealing lwith lthe luncertainty lof lwhat lthe

lcurrent lstate lis, lwhat lthe loutcome lof lpossible lactions lmight lbe, land lwhat lis lit lthat lwe

lreally ldesire.

c. We ldefine lan lagent las lan lentity lthat ltakes laction lin lresponse lto lpercepts lfrom lan lenvi-
lronment.

d. We ldefine lrationality las lthe lproperty lof la lsystem lwhich ldoes lthe l“right lthing” lgiven
lwhat lit lknows. l See lSection l2.2 lfor la lmore lcomplete ldiscussion. l The lbasic lconcept lis

lperfect lrationality; lSection l?? ldescribes lthe limpossibility lof lachieving lperfect lrational-

lity land lproposes lan lalternative ldefinition.

e. We ldefine llogical lreasoning las lthe la lprocess lof lderiving lnew lsentences lfrom lold, lsuch
lthat lthe lnew lsentences lare lnecessarily ltrue lif lthe lold lones lare ltrue. l(Notice lthat ldoes lnot

lrefer lto lany lspecific lsyntax lor lformal llanguage, lbut lit ldoes lrequire la lwell-defined lnotion

lof ltruth.)




Exercise l1.1.#TURI
Read lTuring’s loriginal lpaper lon lAI l(Turing, l1950). l In lthe lpaper, lhe ldiscusses lseveral
lobjections lto lhis lproposed lenterprise land lhis ltest lfor lintelligence. l Which lobjections lstill lcarry




© l2023 lPearson lEducation, lHoboken, lNJ. lAll lrights lreserved.