SOLUTIONS & INSTRUCTOR MANUAL
o o o
Artificial Intelligence: A Modern Approach, 4th Edition
o o o o o o
by Peter Norvig and Stuart Russell, Chapters 1 – 28
o o o o o o o o o o
,Artificial Intelligence
o
oooo 1 Introduction ...
o o o
oooo 2 Intelligent Agents ...
o o o o
II Problem-solving
o
oooo 3 Solving Problems by Searching ...
o o o o o o
oooo 4 Search in Complex Environments ...
o o o o o o
oooo 5 Adversarial Search and Games ...
o o o o o o
oooo 6 Constraint Satisfaction Problems ...
o o o o o
III Knowledge, reasoning, and planning
o o o o
oooo 7 Logical Agents ...
o o o o
oooo 8 First-Order Logic ...
o o o o
oooo 9 Inference in First-Order Logic ...
o o o o o
oooo 10 Knowledge Representation ...
o o o o
oooo 11 Automated Planning ...
o o o o
IV Uncertain knowledge and reasoning
o o o o
oooo 12 Quantifying Uncertainty ...
o o o o
oooo 13 Probabilistic Reasoning ...
o o o o
oooo 14 Probabilistic Reasoning over Time ...
o o o o o o
oooo 15 Probabilistic Programming ...
o o o o
oooo 16 Making Simple Decisions ...
o o o o o
oooo 17 Making Complex Decisions ...
o o o o o
oooo 18 Multiagent Decision Making ...
o o o o o
V Machine Learning
o o
,oooo 19 Learning from Examples ...
o o o o o
oooo 20 Learning Probabilistic Models ...
o o o o o
oooo 21 Deep Learning ...
o o o o
oooo 22 Reinforcement Learning ...
o o o o
VI Communicating, perceiving, and acting
o o o o
oooo 23 Natural Language Processing ...
o o o o o
oooo 24 Deep Learning for Natural Language Processing ...
o o o o o o o o
oooo 25 Computer Vision ...
o o o o
oooo 26 Robotics ...
o o o
VII Conclusions
o
oooo 27 Philosophy, Ethics, and Safety of AI ...
o o o o o o o o
oooo 28 The Future of AI
o o o o
, EXERCISES o o
1
INTRODUCTION
Note othat ofor omany oof othe oquestions oin othis ochapter, owe ogive oreferences owhere oanswers
ocan obe ofound orather othan owriting othem oout—the ofull oanswers owould obe ofar otoo olong.
1.1 What Is AI?
o o o
Exercise o1.1.#DEFA
Define oin oyour oown owords: o (a) ointelligence, o(b) oartificial ointelligence, o(c) oagent, o(d)
ora- otionality, o(e) ological oreasoning.
a. Dictionary odefinitions oof ointelligence otalk oabout o“the ocapacity oto oacquire oand
oapply oknowledge” oor o“the ofaculty oof othought oand oreason” oor o“the oability oto
ocomprehend oand oprofit ofrom oexperience.” o These oare oall oreasonable oanswers, obut
oif owe owant osomething oquantifiable owe owould ouse osomething olike o“the oability oto
oact osuccessfully oacross oa owide orange oof oobjectives oin ocomplex oenvironments.”
b. We odefine oartificial ointelligence oas othe ostudy oand oconstruction oof oagent oprograms
othat operform owell oin oa ogiven oclass oof oenvironments, ofor oa ogiven oagent
oarchitecture; othey odo othe oright othing. o An oimportant opart oof othat ois odealing owith
othe ouncertainty oof owhat othe ocurrent ostate ois, owhat othe ooutcome oof opossible oactions
omight obe, oand owhat ois oit othat owe oreally odesire.
c. We odefine oan oagent oas oan oentity othat otakes oaction oin oresponse oto opercepts ofrom oan
oenvi- oronment.
d. We odefine orationality oas othe oproperty oof oa osystem owhich odoes othe o“right othing”
ogiven owhat oit oknows. o See oSection o2.2 ofor oa omore ocomplete odiscussion. o The
obasic oconcept ois operfect orationality; oSection o?? odescribes othe oimpossibility oof
oachieving operfect orational- oity oand oproposes oan oalternative odefinition.
e. We odefine ological oreasoning oas othe oa oprocess oof oderiving onew osentences ofrom oold,
osuch othat othe onew osentences oare onecessarily otrue oif othe oold oones oare otrue. o(Notice
othat odoes onot orefer oto oany ospecific osyntax oor oformal olanguage, obut oit odoes orequire oa
owell-defined onotion oof otruth.)
Exercise o1.1.#TURI
Read oTuring’s ooriginal opaper oon oAI o(Turing, o1950). o In othe opaper, ohe odiscusses
oseveral oobjections oto ohis oproposed oenterprise oand ohis otest ofor ointelligence. oWhich oobjections
ostill ocarry
© o2023 oPearson oEducation, oHoboken, oNJ. oAll orights
oreserved.
o o o
Artificial Intelligence: A Modern Approach, 4th Edition
o o o o o o
by Peter Norvig and Stuart Russell, Chapters 1 – 28
o o o o o o o o o o
,Artificial Intelligence
o
oooo 1 Introduction ...
o o o
oooo 2 Intelligent Agents ...
o o o o
II Problem-solving
o
oooo 3 Solving Problems by Searching ...
o o o o o o
oooo 4 Search in Complex Environments ...
o o o o o o
oooo 5 Adversarial Search and Games ...
o o o o o o
oooo 6 Constraint Satisfaction Problems ...
o o o o o
III Knowledge, reasoning, and planning
o o o o
oooo 7 Logical Agents ...
o o o o
oooo 8 First-Order Logic ...
o o o o
oooo 9 Inference in First-Order Logic ...
o o o o o
oooo 10 Knowledge Representation ...
o o o o
oooo 11 Automated Planning ...
o o o o
IV Uncertain knowledge and reasoning
o o o o
oooo 12 Quantifying Uncertainty ...
o o o o
oooo 13 Probabilistic Reasoning ...
o o o o
oooo 14 Probabilistic Reasoning over Time ...
o o o o o o
oooo 15 Probabilistic Programming ...
o o o o
oooo 16 Making Simple Decisions ...
o o o o o
oooo 17 Making Complex Decisions ...
o o o o o
oooo 18 Multiagent Decision Making ...
o o o o o
V Machine Learning
o o
,oooo 19 Learning from Examples ...
o o o o o
oooo 20 Learning Probabilistic Models ...
o o o o o
oooo 21 Deep Learning ...
o o o o
oooo 22 Reinforcement Learning ...
o o o o
VI Communicating, perceiving, and acting
o o o o
oooo 23 Natural Language Processing ...
o o o o o
oooo 24 Deep Learning for Natural Language Processing ...
o o o o o o o o
oooo 25 Computer Vision ...
o o o o
oooo 26 Robotics ...
o o o
VII Conclusions
o
oooo 27 Philosophy, Ethics, and Safety of AI ...
o o o o o o o o
oooo 28 The Future of AI
o o o o
, EXERCISES o o
1
INTRODUCTION
Note othat ofor omany oof othe oquestions oin othis ochapter, owe ogive oreferences owhere oanswers
ocan obe ofound orather othan owriting othem oout—the ofull oanswers owould obe ofar otoo olong.
1.1 What Is AI?
o o o
Exercise o1.1.#DEFA
Define oin oyour oown owords: o (a) ointelligence, o(b) oartificial ointelligence, o(c) oagent, o(d)
ora- otionality, o(e) ological oreasoning.
a. Dictionary odefinitions oof ointelligence otalk oabout o“the ocapacity oto oacquire oand
oapply oknowledge” oor o“the ofaculty oof othought oand oreason” oor o“the oability oto
ocomprehend oand oprofit ofrom oexperience.” o These oare oall oreasonable oanswers, obut
oif owe owant osomething oquantifiable owe owould ouse osomething olike o“the oability oto
oact osuccessfully oacross oa owide orange oof oobjectives oin ocomplex oenvironments.”
b. We odefine oartificial ointelligence oas othe ostudy oand oconstruction oof oagent oprograms
othat operform owell oin oa ogiven oclass oof oenvironments, ofor oa ogiven oagent
oarchitecture; othey odo othe oright othing. o An oimportant opart oof othat ois odealing owith
othe ouncertainty oof owhat othe ocurrent ostate ois, owhat othe ooutcome oof opossible oactions
omight obe, oand owhat ois oit othat owe oreally odesire.
c. We odefine oan oagent oas oan oentity othat otakes oaction oin oresponse oto opercepts ofrom oan
oenvi- oronment.
d. We odefine orationality oas othe oproperty oof oa osystem owhich odoes othe o“right othing”
ogiven owhat oit oknows. o See oSection o2.2 ofor oa omore ocomplete odiscussion. o The
obasic oconcept ois operfect orationality; oSection o?? odescribes othe oimpossibility oof
oachieving operfect orational- oity oand oproposes oan oalternative odefinition.
e. We odefine ological oreasoning oas othe oa oprocess oof oderiving onew osentences ofrom oold,
osuch othat othe onew osentences oare onecessarily otrue oif othe oold oones oare otrue. o(Notice
othat odoes onot orefer oto oany ospecific osyntax oor oformal olanguage, obut oit odoes orequire oa
owell-defined onotion oof otruth.)
Exercise o1.1.#TURI
Read oTuring’s ooriginal opaper oon oAI o(Turing, o1950). o In othe opaper, ohe odiscusses
oseveral oobjections oto ohis oproposed oenterprise oand ohis otest ofor ointelligence. oWhich oobjections
ostill ocarry
© o2023 oPearson oEducation, oHoboken, oNJ. oAll orights
oreserved.
