SOLUTIONS & INSTRUCTOR MANUAL
nii nii nii
Artificial Intelligence: A Modern Approach, 4th Edition
nii nii nii nii nii nii
by Peter Norvig and Stuart Russell, Chapters 1 – 28
nii nii nii nii nii nii nii nii nii
,Artificial Intelligence
n
1 Introduction ... n
2 Intelligent Agents ... n n
II Problem-solving
n
3 Solving Problems by Searching ...
n n n n
4 Search in Complex Environments ...
n n n n
5 Adversarial Search and Games ... n n n n
6 Constraint Satisfaction Problems ... n n n n
III Knowledge, reasoning, and planning
n n n n
7 Logical Agents ...
n n
8 First-Order Logic ... n n
9 Inference in First-Order Logic ...
n n n n
10 Knowledge Representation ... n n
11 Automated Planning ... n n
IV Uncertain knowledge and reasoning
n n n n
12 Quantifying Uncertainty ... n n
13 Probabilistic Reasoning ... n n
14 Probabilistic Reasoning over Time ... n n n n
15 Probabilistic Programming ... n n
16 Making Simple Decisions ...
n n n
17 Making Complex Decisions ...
n n n
18 Multiagent Decision Making ... n n n
V Machine Learning
n n
, 19 Learning from Examples ...
n n n
20 Learning Probabilistic Models ...
n n n
21 Deep Learning ...
n n
22 Reinforcement Learning ... n n
VI Communicating, perceiving, and acting
n n n n
23 Natural Language Processing ...
n n n
24 Deep Learning for Natural Language Processing ...
n n n n n n
25 Computer Vision ... n n
26 Robotics ... n n
VII Conclusions
n
27 Philosophy, Ethics, and Safety of AI ...
n n n n n n
28 The Future of AI
n n n
, EXERCISES n in in
1
INTRODUCTION
Noten thatn forn manyn ofn then questionsn inn thisn chapter,n wen given referencesn wheren answersnca
n n ben found n rathern than n writingn themn out—then fulln answersn would n ben farn too n long.
1.1 What Is AI?n n
Exercise ii1.1.#DEFA
Defineniiin niiyourniiown niiwords:n inin (a)niiintelligence,nii(b)niiartificialniiintelligence,nii(c)niiagent,nii(d)n
iira-niitionality, nii(e)niilogicalniireasoning.
a. Dictionaryn definitionsn ofn intelligencen talkn aboutn “then capacityn ton acquiren andn appl
y nknowledge”norn “thenfacultyn ofn thoughtn and nreason”norn “then abilityn to ncomprehend n an
d n profitn fromn experience.”n n Thesen aren alln reasonable n answers,nbutn ifn wen wantn
somethingn quantifiablen wen wouldn usen somethingn liken “then abilityn to n actn successfull
y n acrossn an widen rangen of n objectivesn in n complex n environments.”
b. Wen definen artificialn intelligencen asn then studyn andn constructionn ofn agentnprogram
sn thatn performn welln in n an givenn classn of n environments, n forn an given nagentn arc
hitecture;n they n don then rightn thing.n n An n importantn partn of n thatn isndealingn with n
then uncertainty nofn whatn then currentn staten is,n whatn then outcomenof n possiblen actionsn mi
ghtn be,n and n whatn isn itn thatn wen really n desire.
c. Wen definen ann agentn asn ann entityn thatn takesn actionn inn responsen ton perceptsn fromnan
n envi-n ronment.
d. Wen definen rationalityn asn then propertyn ofn an systemn whichn doesn then “rightn thing”n g
ivenn whatn itn knows.n n Seen Sectionn 2.2n forn an moren completen discussion.n n Thenbasicn
conceptn isnperfectn rationality;n Sectionn ?? ndescribesn then impossibility nofn achievingnperf
ectn rational-n ity n and n proposesn an n alternative n definition.
e. Wendefinen logicalnreasoning nasn thenanprocessnof nderivingnnewnsentencesnfromnold,nsuc
hn thatn thennewn sentencesnarennecessarilyn truen ifnthenold nonesnaren true.n (Noticen thatndoesn
notn refern to nanyn specificn syntax nornformaln language,nbutn itn doesn requiren an well-
n defined n notion n of n truth.)
Exercise ii1.1.#TURI
Read niiTuring’sniioriginalniipaperniion niiAInii(Turing,nii1950).n i nin Inniitheniipaper,niiheniidiscussesniise
veralniiobjectionsniitoniihisniiproposedniienterpriseniiandniihisniitestniiforniiintelligence.niiWhichiiobjectionsniistillnii
carry
©n 2023n Pearsonn Education,n Hoboken,n NJ.n All n right
s
nii nii nii
Artificial Intelligence: A Modern Approach, 4th Edition
nii nii nii nii nii nii
by Peter Norvig and Stuart Russell, Chapters 1 – 28
nii nii nii nii nii nii nii nii nii
,Artificial Intelligence
n
1 Introduction ... n
2 Intelligent Agents ... n n
II Problem-solving
n
3 Solving Problems by Searching ...
n n n n
4 Search in Complex Environments ...
n n n n
5 Adversarial Search and Games ... n n n n
6 Constraint Satisfaction Problems ... n n n n
III Knowledge, reasoning, and planning
n n n n
7 Logical Agents ...
n n
8 First-Order Logic ... n n
9 Inference in First-Order Logic ...
n n n n
10 Knowledge Representation ... n n
11 Automated Planning ... n n
IV Uncertain knowledge and reasoning
n n n n
12 Quantifying Uncertainty ... n n
13 Probabilistic Reasoning ... n n
14 Probabilistic Reasoning over Time ... n n n n
15 Probabilistic Programming ... n n
16 Making Simple Decisions ...
n n n
17 Making Complex Decisions ...
n n n
18 Multiagent Decision Making ... n n n
V Machine Learning
n n
, 19 Learning from Examples ...
n n n
20 Learning Probabilistic Models ...
n n n
21 Deep Learning ...
n n
22 Reinforcement Learning ... n n
VI Communicating, perceiving, and acting
n n n n
23 Natural Language Processing ...
n n n
24 Deep Learning for Natural Language Processing ...
n n n n n n
25 Computer Vision ... n n
26 Robotics ... n n
VII Conclusions
n
27 Philosophy, Ethics, and Safety of AI ...
n n n n n n
28 The Future of AI
n n n
, EXERCISES n in in
1
INTRODUCTION
Noten thatn forn manyn ofn then questionsn inn thisn chapter,n wen given referencesn wheren answersnca
n n ben found n rathern than n writingn themn out—then fulln answersn would n ben farn too n long.
1.1 What Is AI?n n
Exercise ii1.1.#DEFA
Defineniiin niiyourniiown niiwords:n inin (a)niiintelligence,nii(b)niiartificialniiintelligence,nii(c)niiagent,nii(d)n
iira-niitionality, nii(e)niilogicalniireasoning.
a. Dictionaryn definitionsn ofn intelligencen talkn aboutn “then capacityn ton acquiren andn appl
y nknowledge”norn “thenfacultyn ofn thoughtn and nreason”norn “then abilityn to ncomprehend n an
d n profitn fromn experience.”n n Thesen aren alln reasonable n answers,nbutn ifn wen wantn
somethingn quantifiablen wen wouldn usen somethingn liken “then abilityn to n actn successfull
y n acrossn an widen rangen of n objectivesn in n complex n environments.”
b. Wen definen artificialn intelligencen asn then studyn andn constructionn ofn agentnprogram
sn thatn performn welln in n an givenn classn of n environments, n forn an given nagentn arc
hitecture;n they n don then rightn thing.n n An n importantn partn of n thatn isndealingn with n
then uncertainty nofn whatn then currentn staten is,n whatn then outcomenof n possiblen actionsn mi
ghtn be,n and n whatn isn itn thatn wen really n desire.
c. Wen definen ann agentn asn ann entityn thatn takesn actionn inn responsen ton perceptsn fromnan
n envi-n ronment.
d. Wen definen rationalityn asn then propertyn ofn an systemn whichn doesn then “rightn thing”n g
ivenn whatn itn knows.n n Seen Sectionn 2.2n forn an moren completen discussion.n n Thenbasicn
conceptn isnperfectn rationality;n Sectionn ?? ndescribesn then impossibility nofn achievingnperf
ectn rational-n ity n and n proposesn an n alternative n definition.
e. Wendefinen logicalnreasoning nasn thenanprocessnof nderivingnnewnsentencesnfromnold,nsuc
hn thatn thennewn sentencesnarennecessarilyn truen ifnthenold nonesnaren true.n (Noticen thatndoesn
notn refern to nanyn specificn syntax nornformaln language,nbutn itn doesn requiren an well-
n defined n notion n of n truth.)
Exercise ii1.1.#TURI
Read niiTuring’sniioriginalniipaperniion niiAInii(Turing,nii1950).n i nin Inniitheniipaper,niiheniidiscussesniise
veralniiobjectionsniitoniihisniiproposedniienterpriseniiandniihisniitestniiforniiintelligence.niiWhichiiobjectionsniistillnii
carry
©n 2023n Pearsonn Education,n Hoboken,n NJ.n All n right
s
