AIMA 3ed. - Solution Manual

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Instructor’s Manual: Exercise Solutions for Artificial Intelligence AModernApproach Third Edition (International Version) Stuart J. Russell and Peter Norvig with contributions from Ernest Davis, Nicholas J. Hay, and Mehran Sahami Upper Saddle River Boston Columbus San Francisco New York Indianapolis London Toronto Sydney Singapore Tokyo Montreal Dubai Madrid Hong Kong Mexico City Munich Paris Amsterdam Cape Town

• Page 2 and 3: Editor-in-Chief: Michael Hirsch Exe
• Page 5 and 6: Solutions for Chapter 1 Introductio
• Page 7 and 8: 3 The voice activated directory ass
• Page 9 and 10: 5 d. (shoppingatthemarket)No.Norobo
• Page 11 and 12: 7 The dominant approach has switche
• Page 13 and 14: 9 EXTERNAL MEMORY was already clean
• Page 15 and 16: 11 • Rationality: apropertyofagen
• Page 17 and 18: 13 function UTILITY-BASED-AGENT(per
• Page 19 and 20: 15 update is nontrivial to explain.
• Page 21 and 22: 17 3.3 predicting the outcome of ou
• Page 23 and 24: 19 3.7 3.8 c. Initialstate:consider
• Page 25 and 26: 21 how the world is, because the ag
• Page 27 and 28: 23 3.16 d. Yes;startatthegoal,andap
• Page 29 and 30: 25 C[265+160=425], T[111+329=440],
• Page 31 and 32: 27 case, assume n ′ is on the sho
• Page 33 and 34: 29 and hole on already-linked piece
• Page 35 and 36: 31 function AND-OR-GRAPH-SEARCH(pro
• Page 37 and 38: 33 L R L R S S S Figure S4.3 The be
• Page 39 and 40: 35 Current position Current positio
• Page 41 and 42: 37 bd −4 cd −4 ad
• Page 43 and 44: 39 MAX A a 1 a 2 MIN B D b1 b2 b3 d
• Page 45 and 46: 41 1 x x −1 x 1 −2 x o x x o x
• Page 47 and 48: 43 1.5 1.5 −0.5 0.5 0.5 0.5 0.5 2
• Page 49 and 50: 45 b. In a partially observable, tu
• Page 51 and 52: 47 that the letters must make words
• Page 53 and 54: 49 f. A 2 = R conflicts with A 1 ,
• Page 55 and 56: Solutions for Chapter 7 Logical Age
• Page 57 and 58: 53 function PL-TRUE?(s, m) returns
• Page 59 and 60: 55 7.8 Abinarylogicalconnectiveisde
• Page 61 and 62: 57 •ResolveS7withS5,givingS8:F .
• Page 63 and 64: 59 7.18 7.19 clauses are semantical
• Page 65 and 66: 61 aMinesweepergamewith100cellsand2
• Page 67 and 68: Solutions for Chapter 8 First-Order
• Page 69 and 70: Page 71 and 72: 67 8.10 (iv) ∀ x, y (Country(x)
• Page 73 and 74: 69 example, it is not possible to p
• Page 75 and 76: 71 less felicitous); Sells(x, y, z)
• Page 77 and 78: 73 connections: ∀ c Add 1 Circuit
• Page 79 and 80: Solutions for Chapter 9 Inference i
• Page 81 and 82: 77 a. LetP (x, y) be the relation
• Page 83 and 84: 79 (and negating Q and including th
• Page 85 and 86: 81 Horse(h) Offspring(h,y) Horse(Bl
• Page 87 and 88: 83 a. Thecodeforsimplificationlooks
• Page 89 and 90: 85 Note that this does not work in
• Page 91 and 92: Solutions for Chapter 10 Classical
• Page 93 and 94: 89 The initial state is: In(Switch
• Page 95 and 96: 91 can still fly when empty. (2) Ne
• Page 97 and 98: a. Yes,thiswillfindaplanwheneverthe
• Page 99 and 100: 95 Refinement(Deliver(t, p), PRECON
• Page 101 and 102: 97 11.8 Flip can be described using
• Page 103 and 104: 99 [q1=Q11 ∧ q2=Q12 ∧ q3=Q13]
• Page 105 and 106: 101 use the first sense, so we will
• Page 107 and 108: 103 There is no obvious way to refe
• Page 109 and 110: 105 12.16 Let Trade(b, x, a, y) den
• Page 111 and 112: 107 d. Ifeachcategoryhasmanypropert
• Page 113 and 114: 109 We then have the following equa
• Page 115 and 116: 111 d. Thisasksforthevectorofprobab
• Page 117 and 118: 113 test(10000) 13.11 The correct m
• Page 119 and 120: 115 13.16 The basic axiom to use he
• Page 121 and 122: 117 Normalization proceeds as follo
• Page 123 and 124: 119 To calculate the probabilities
• Page 125 and 126: Solutions for Chapter 14 Probabilis
• Page 127 and 128: 123 14.3 This generalizes the sum-t
• Page 129 and 130: 125 a. Yes. Numerically one can com
• Page 131 and 132: 127 or a parent. Hence, P (x i |x 1
• Page 133 and 134: 129 have P (x) = “ 1 √ 1 (2π)
• Page 135 and 136: 131 T = Normal T = High d. TheCPTfo
• Page 137 and 138: 133 e. Wecannotcalculatethemostlike
• Page 139 and 140: 135 14.16 leaf node, we have to do
• Page 141 and 142: 137 •Entriesonthediagonalcorrespo
• Page 143 and 144: 139 A.Q B.Q C.Q AB.Outcome BC.Outco
• Page 145 and 146: 141 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0
• Page 147 and 148: 143 Localization error 6 5 4 3 2 1
• Page 149 and 150: 145 7 Variance of posterior distrib
• Page 151 and 152: 147 = ⟨0.02, 0.21⟩ P (e 3 |S 2
• Page 153 and 154: Solutions for Chapter 16 Making Sim
• Page 155 and 156: 151 × P(Wrapper = red|Flavor = anc
• Page 157 and 158: 153 Solving this numerically, we fi
• Page 159 and 160: 155 Using Bayes’ theorem: P (q +
• Page 161 and 162: Solutions for Chapter 17 Making Com
• Page 163 and 164: 159 One useful observation in this
• Page 165 and 166: 161 (intersection-point (make-line
• Page 167 and 168: 163 a. ForU A we have U A (s)=R(s)+
• Page 169 and 170: 165 17.10 a. Intuitively,theagentwa
• Page 171 and 172: 167 Given this policy, the policy l
• Page 173 and 174: 169 for the item. Now, agent i wins
• Page 175 and 176: 171 17.21 Every game is either a wi
• Page 177 and 178: 173 only positive examples, then th
• Page 179 and 180: 175 18.6 Note that to compute each
• Page 181 and 182: 177 decision list in which C i is t
• Page 183 and 184: 179 18.21 This question introduces
• Page 185 and 186: 181 c. Fourpointsingeneralpositiono
• Page 187 and 188: 183 That is, if p has the same inpu
• Page 189 and 190: 185 Again, we have a fairly free ch
• Page 191 and 192: Solutions for Chapter 20 Learning P
• Page 193 and 194: 189 parameter estimation based on c
• Page 195 and 196: = Γ(a + b) Γ(a)Γ(b) · aΓ(a)Γ(
• Page 197 and 198: 193 and the 90 green-wrapped cherry
• Page 199 and 200: 195 zero under these conditions, so
• Page 201 and 202: Solutions for Chapter 21 Reinforcem
• Page 203 and 204: 199 part of the learning algorithm
• Page 205 and 206: 201 return None return words # for
• Page 207 and 208: Solutions for Chapter 23 Natural La
• Page 209 and 210: 205 a. Thelanguagea n b n :Thestrin
• Page 211 and 212: 207 S → NP VP | S ′ Conj S ′
• Page 213 and 214: 209 As these are the only two parse
• Page 215 and 216: 211 •Puttingtoomanyclothestogethe
• Page 217 and 218: 213 θ r x illumination y z x viewe
• Page 219 and 220: Solutions for Chapter 25 Robotics 2
• Page 221 and 222: 217 Aquickcheckshouldconvinceyoutha
• Page 223 and 224: 219 value of the term “1/square o
• Page 225 and 226: 221 Asingleplanarobstacleprotruding
• Page 227 and 228: 223 Acommonreactivealgorithm,whichh
• Page 229 and 230: aby. The instructor then follows th
• Page 231 and 232: 227 h. make mistakes At this stage,
• Page 233 and 234: 229 getting a “human” answer, n
• Page 235 and 236: Bibliography 231
• Page 237: Bibliography 233 Tesauro, G. (1992)