instructor s solutions manual artificial intelligence a modern approach
LINK 1 ENTER SITE >>> Download PDF
LINK 2 ENTER SITE >>> Download PDF
File Name:instructor s solutions manual artificial intelligence a modern approach.pdf
Size: 3752 KB
Type: PDF, ePub, eBook
Category: Book
Uploaded: 23 May 2019, 13:56 PM
Rating: 4.6/5 from 828 votes.
Status: AVAILABLE
Last checked: 15 Minutes ago!
In order to read or download instructor s solutions manual artificial intelligence a modern approach ebook, you need to create a FREE account.
eBook includes PDF, ePub and Kindle version
✔ Register a free 1 month Trial Account.
✔ Download as many books as you like (Personal use)
✔ Cancel the membership at any time if not satisfied.
✔ Join Over 80000 Happy Readers
instructor s solutions manual artificial intelligence a modern approachSome features of WorldCat will not be available.By continuing to use the site, you are agreeing to OCLC’s placement of cookies on your device. Find out more here. However, formatting rules can vary widely between applications and fields of interest or study. The specific requirements or preferences of your reviewing publisher, classroom teacher, institution or organization should be applied. Please enter recipient e-mail address(es). Please re-enter recipient e-mail address(es). Please enter your name. Please enter the subject. Please enter the message. Author: Stuart J Russell; Peter NorvigPlease select Ok if you would like to proceed with this request anyway. All rights reserved. You can easily create a free account. Please try again.Please try again.Please try again. Then you can start reading Kindle books on your smartphone, tablet, or computer - no Kindle device required. To calculate the overall star rating and percentage breakdown by star, we don’t use a simple average. Instead, our system considers things like how recent a review is and if the reviewer bought the item on Amazon. It also analyzes reviews to verify trustworthiness. Please try again later. Sherry 1.0 out of 5 stars. Page Count: 181 Exercise SolutionsArtificial Intelligence. A Modern Approach. Second Edition. Stuart J. Russell and Peter Norvig. Upper Saddle River, New Jersey 07458Russell, Stuart J. (Stuart Jonathan). Instructor’s solution manual for artificial intelligence: a modern approach. Includes bibliographical references and index.Vice President and Editorial Director, ECS: Marcia J. Horton. Publisher: Alan R. Apt. Associate Editor: Toni Dianne Holm. Editorial Assistant: Patrick Lindner. Vice President and Director of Production and Manufacturing, ESM: David W. Riccardi. Executive Managing Editor: Vince O’Brien. Managing Editor: Camille Trentacoste. Production Editor: Mary Massey. Manufacturing Manager: Trudy Pisciotti. Manufacturing Buyer: Lisa McDowell.http://cityini.com/files/fckeditor/ic-a22-service-manual.xml
- Tags:
- instructor s solutions manual artificial intelligence a modern approach.
Marketing Manager: Pamela ShafferPearson Prentice Hall. Pearson Education, Inc. Upper Saddle River, NJ 07458. All rights reserved. No part of this manual may be reproduced in any form or by any means, withoutPearson Prentice Hall R is a trademark of Pearson Education, Inc. Printed in the United States of AmericaPearson Education Australia Pty. Ltd., Sydney. Pearson Education Singapore, Pte. Ltd. Pearson Education North Asia Ltd., Hong Kong. Pearson Education Canada, Inc., Toronto. Pearson Educacio?n de Mexico, S.A. de C.V. Pearson Education—Japan, Tokyo. Pearson Education Malaysia, Pte. Ltd. Pearson Education, Inc., Upper Saddle River, New JerseyThis Instructor’s Solution Manual provides solutions (or at least solution sketches) forWe only give actual code for a few of the programming exercises; writing a lot of codeIn many cases, we give ideas for discussion and follow-up questions, and we try toThere is more supplementary material that we want to offer to the instructor, but weThe idea is that this solution manual contains the material thatThe address for the web site is:Instructions on how to join the aima-instructors discussion list. We strongly recommend that you join so that you can receive updates, corrections, notification of newSource code for programs from the text. We offer code in Lisp, Python, and Java, and. Programming resources and supplemental texts. Figures from the text; for overhead transparencies. Terminology from the index of the book. Other courses using the book that have home pages on the Web. You can see examplePlease do not put solution sets for AIMA exercises onAI Education information on teaching introductory AI courses. Other sites on the Web with information on AI. Organized by chapter in the book; checkWe welcome suggestions for new exercises, new environments and agents, etc. TheWe hope that you enjoy teaching from it,IntroductionThe probability of fooling an interrogator depends on just how unskilled the interrogator is.http://www.xn----qtbenjffc7h.xn--p1ai/userfiles/ic-a3e-manual.xml One entrant in the 2002 Loebner prize competition (which is not quite a real Turing. Test) did fool one judge, although if you look at the transcript, it is hard to imagine whatThere certainly have been examples of a chatbot or other onlineFor example, see See Lenny Foner’s account of the Julia chatbot. We’d say the chance today is somethingIt is providing evidence on the number and type of rules that are sufficient for producing one type ofNotice that humans don’t solve NPcomplete problems either. Sometimes they are good at solving specific instances with a lot ofIntroductionAI systems should attempt to doThe IQ test doesn’t measure everything. A program that is specialized only for IQ tests (and specialized further only for the analogySee The Mismeasure of Man by Stephen Jay Gould, Norton, 1981 or Multiple intelligences: the theory inYou do have a conscious awarenessThe field of psychoanalysis is based on the idea that one needs trained professional help toAlthough there has been a lot of progress in automated driving,To our knowledge, none are able to avoidDriving in downtown Cairo is too unpredictableNo robot can currently put together the tasks of moving inThe componentSoftware robots are capable of handling such tasks, particularly if the design of the web grocery shopping site does not change radically overFor example, the proof of Robbins algebra described on pageAI has a long history of research into applicationsTwo outstanding examples are the Prolog-based expertThe social security system is said to have savedSee Kay, Gawron and. Norvig (1994) and Wahlster (2000) for an overview of the field of speech translation,But given a percept, an agent still has the task of “deciding” (either by deliberation or byThis is just as true in the real world as in artificial microworlds such as chess-playing.https://www.informaquiz.it/petrgenis1604790/status/flotaganis19052022-1933 So computing the appropriate action will remain a crucial partOn the other hand, it is true that a concentration on micro-worlds has led AI away from theThat is, evolution favors organisms that canRationality just means optimizing performance measure, so this is inIf “intelligent” means “applying knowledge” or “using thoughtSo in one sense Samuel “told” theWhatever youIntroductionNote that Searle makes this appeal to mechanismIntelligent AgentsAgent: an entity that perceives and acts; or, one that can be viewed as perceiving andAgent function: a function that specifies the agent’s action in response to every possibleAgent program: that program which, combined with a machine architecture, implements an agent function. In our simple designs, the program takes a new percept onRationality: a property of agents that choose actions that maximize their expected utility, given the percepts to date. Autonomy: a property of agents whose behavior is determined by their own experienceReflex agent: an agent whose action depends only on the current percept. Model-based agent: an agent whose action is derived directly from an internal modelGoal-based agent: an agent that selects actions that it believes will achieve explicitlyUtility-based agent: an agent that selects actions that it believes will maximize theLearning agent: an agent whose behavior improves over time based on its experience.A utility function is used by an agentIn our framework, the utility functionOur answers are for the simple agent designs for static environments where nothing happensWhen there is dirt in the initial location and none inIn this particular case, however, thatIf one of the three isThus the change in the sumSo before we solve a puzzle, we should compute the P value of the start and goal stateConsider a stateThis yields an infiniteHowever, if the stateGoal test: All regions colored, and no two adjacent regions have the same color. Successor function: Assign a color to a region.http://dieter-sauter.com/images/canon-laser-class-710-fax-machine-owners-manual.pdf Cost function: Number of assignments.Goal test: Monkey has bananas. Successor function: Hop on crate; Hop off crate; Push crate from one spot to another. Walk from one spot to another; grab bananas (if standing on crate). Cost function: Number of actions.Goal test: considering a single record, and it gives “illegal input” message. Successor function: run again on the first half of the records; run again on the secondCost function: Number of runs. Note: This is a contingency problem; you need to see whether a run gives an errorBacktracking search is a form depth-first search in which there is a single representation of the state that gets updated for each successor, and then must be restored when a deadBackjumping is a way of making backtracking search more efficient, by jumping backMin-conflicts is a heuristic for use with local search on CSP problems. The heuristicStart with SA, whichThen moving clockwise, WA can have either of the other twoThe least constraining valueOne simple choice isYou could have a variable for each box inThis approach is feasible with a most-constraining valueBoth formulations areConstraints say that noThis is very similar to the forwardTo express the ternaryAll other ternary constraints can be handled similarly. Now that we can reduce a ternary constraint into binary constraints, we can reduce aWhether this makes the cycle cutset approach practical depends more on the graphSo any graph with a largeThe value of each variableThis is a good representation because itAnother representation is to have five variables for each house, one with the domain ofAdversarial SearchThe values imply thatThe terminal nodes with a bold outline areFigure S6.1If MIN plays suboptimally,For example, if MIN always falls for a certain kind of trap and loses, then setting the trapThis is shown in. Figure S6.2.Adversarial SearchThe minimax move is of course o ?, with valueThe game tree for the four-square game in Exercise 6.3. Terminal states areEach state is annotated with its minimax valueIt can be fixedAlthough it works in thisFinally, in games with chance nodes, it is unclear how toWhat is really happening is that each state has a well-defined but initially unknownIf the game tree has cycles, then a dynamic programmingIn such a case, the rules of the game will need to defineIn chess, for eaxmple, a state that occursOne approach is a proof by induction on the size of the. Similarly for uBU.The Java code did not have this function as of May 2003, butThis shows that in general oneEnumerating them takes exponential time.It will do this even if other. Thus, in the partial model whereIt would be slowerThat is precisely theRemember to count the propositions that are not mentioned; if a sentence mentions only 5 andEach of the fourOne of them is reverse implication ( g instead ofASCII. To save space in this manual, we only show the first four truth tables:Many people are fooled by (e) and (g) because they think of implication as being causation, or something close to it. Thus, in (e), they feel that it is the combination of SmokeHowever, this reasoning isThat means itThe first disjunct isHFEdUE-q EThere will be !a disjuncts, each saying that.I disjuncts. Conjoin all the sentences together. Then use DPLL to answer the question of whether this sentence entails U a y for theTo encode the global constraint that there are s mines altogether, we can constructRemember, uwvHowever, we can represent the globalWe add the parameter min and max toFor an unconstrained problem the values 0 andFor a mineseeper problem the value s will beWithin DPLL, we fail (return false) immediately if min isNo conclusions are invalidated by adding this capability to DPLL and encoding theConsider this string of alternating 1’s and unprobed cells (indicated by a dash). There are two possible models: either there are mines under every even-numberedMaking a probe at either end will determineS UBST z(?Chapter 10 describes the semantics andNOTE: Avoid assigning this problem if you don’t feel comfortable requiring studentsFirst-Order LogicThe key distinctionExplicit sentences: thereImplicit sentences: Van NessFacts such asIn this case, the advantage of the map is really in the ease ofAnalog audio tape recording. Advantages: simple circuits can record and reproduce sounds. Disadvantages: subject to errors, noise; hard to process in order toTraditional clock face. Advantages: easier to read quickly, determination of howDisadvantages: hard toAll kinds of graphs, bar charts, pie charts. Advantages: enormous data compression, easy trend analysis, communicate information in a way which we can interpret easily. Disadvantages: imprecise, cannot represent disjunctive or negatedAccording to the standard semanticsIf there are two objects, thenSome students may also notice that any unsatisfiable sentence alsoEach predicate of arity ? isTwo things to note: first, the number is finite; second, the maximum arityThe main point is to make sure students understand con-UK citizen by birth.J e. That is, a square is breezy if and only if there is a pit in a neighboring square. GenerallyJoeQ ). RoughlyThis argumentThese axioms mayIt allows one to prove, say,If we are using a logic programming system,For example. For the one-bit adder, we haveThe verification query is. It is not possible to ask whether particular terminals are connected in a given circuit, sinceThe two key rules for UK passports are givenInference in First-Order Logic. I?ebeI?ebeBut replacing with any ground term e must count as one of the interpretations, so if theI ? e. U5eS UBST e?iEverest. Note that c does not imply that there are two mountains as high as Everest, because nowhere is it stated that BenNevis is different from Kilimanjaro (or Everest, for thatN?o4o iAlso note that indexing onBut supposeWe unify the. Now take the queryWe then substitute this in to the left-hand side to getWhen we then try to apply the implication again, weIn other words, the failure to standardizeWe want to rewrite this as a single define clause of the formWe can do that with the definite clauseWe use a very simple ontology to make the examples easier:N,u -o-?Whenever a “looping” goalContinue with all other branches of the proofThen use those solutions (suitably instantiated if necessary) as solutions for the suspended subgoal, continuing that branchIn the proof shown in the figure, theSince no otherIn this case, Smith’sOffspring(h,y). Horse(Bluebeard). Parent(y,h). Offspring(Bluebeard,y). Parent(y,Bluebeard). Figure S9.1. Partial proof tree for finding horses.We will useI oN,oGEWe want to be able to show thatThe relevant definitionsI oI oI oE aImplementing the reasoningAlthough the use of logical reasoning for theI u?oE?aThe above axiom for location is extended as follows (note that we do not say whatI u?oAo;?Ao;?Future printings may omit or replace this exercise.One sense of the word “water” includes ice (“that’s frozen water”),The sentences here seem toIt is the sense that is roughly synonymous withThe other tricky part is that we are dealing with objects that change (freeze and melt)For simplicity, we will use a situation calculusThe key thing is to be consistent in the way that information isN I oN I oThe basic meaning of boiling is that instances of the substanceN I o. E E -E?aThen we need only say ?We will use the constant. Note that it is easy to make mistakes in which one asserts that only some of the waterKlcgUKlcgUI,a -Presumably what this means is that all substances that are liquid at room tempero WoIf we use lyNote that this statewhereI,aN I oI oAn interesting exercise would be to define a “pure”I9e oKnowledge RepresentationE nEIt is not the case that sft sftIn that case. We need to ensure that theIf is the set of tomatoes in a bag, thenThen we haveI -oI9e oI -o. I PN I oN ?oWoN ?o. Wo. N ?oSince the conversion axiom for dollars and cents hasIn the new scheme, we must introduce objects whose lengths are converted:It returns a number representing the exchangeThis was the Federal. Reserve bank’s Spot exchange rate as of 10:00 AM. It is the mid-point between the buying andNote also the distinction between a currency such asUsingN o?? E ?EE PE?aE PE ? a?a I?I EOn the other hand,One could also ask students to prove two versions of de Morgan’s laws using the twoI N,oG?For example,To say that an event is fixed is to say that any twoN,y?o oI E ?uWo e o oI oE a ?I E E?aWe can represent requirements as a relation between an individualWe also need to know that a particular object is compatible, i.e., fills a given role appropriately. For example,I E WE?aI?The outcome of this process is a definition of what objects are acceptable to the userLet the termWe want to say about it that transfers the money to, and transfers ownership of to.U yE aNow the only tricky part about defining buying in terms of trading is in distinguishing a priceThe idea is for the students toSome of the key points are:There can be joint ownership and corporate ownership. This suggests the owner is aOwnership provides certain rights: to use, to resell, to give away, etc. Much of this isOwn can own abstract obligations as well as concrete objects. This is the idea behindThis is tricky in terms ofI o a uFor each of these, there may be requirements for the number of courses, the number of units (since different courses may carryWe show our chosen vocabulary byCS1, CS2, CS3, CS21, CS33 and CS34, and some other courses outside the major.Jones meets the requirements for a major in Computer ScienceI u?o a ? -E a E?UE a EE a EI u?o a ?o IOne can easily imagine other kinds of requirements; these just give you a flavor. In this solution we took “over an extended period” to mean that we should recommendBut how do you decide which is bestChapter 16 uses utility theory to address this. IfComplexity is a further problem: witha generalpurpose theorem-prover it’s hard to do much more than enumerate legal programs and pickIn general, students find classification hierarchies easier than other representation tasks. A recent twist is to compare one’s hierarchy with online ones such asIt is suitable for a group project ofThat way, the students see whether they haveThis is true in the logical but not the practicalWe actually would need many more clauses onTo find the value of JB, given a data base with year(jb,1973), make(jb,dodge)I iWithout the distinction beandPlanningBut in planning we open up the representation ofThe applicable actions are:When we add ?Bt. In general, we (1) created a sFaFaE?aN,uN u ? EN,uE i0E -EIt is similarN,oN,o. E?a aN,iAE?aI yI N,oG?I N,oG?I N,oG? I yN,oI N,oG? ACTION: oN,oP RECOND: uN,oN,oI yN,o. N,oN,o. I yI N,oG?N,oN,o. E?aSo there is no way to represent this goal.So not mentioning a literal is the sameSo a negative effect can onlyTherefore, eliminating all negative effects only makes a problem easier.However, most of those who have tried have concluded that biderectional search isA few planners, such as P RODIGY (Fink and. Blythe, 1998) have used bidirectional search.P RODIGY is in fact (in part) a partial-orderThe algorithm does search forward, butFor example, if has three preconditions that can be satisfiedThe level cost (theA backward state-space planner maintains a partial plan that is a reversed sequence ofEYaFirst we’ll explain why it is an anomaly for a noninterleaved planner. There are two subgoals;Now we’ll show how things work out with an interleaved planner such as POP. SinceN e o-?N e oN e oWe then notice that thereN e oN e oN e o-?One solution found by G RAPHPLAN is to execute RightSock and LeftSock in the first timeNow we add the following two actions (neither of which has preconditions):Each of these plans has four steps, and thus fiveThen the final step, Coat, can go inRight. Sock. Right. Shoe. CoatFigure S11.1. Partial-order plan including a hat and coat, for Exercise 11.1.I yI yE FFECT: uI yE FFECT: uACTION:y. P RECOND: uE FFECT:yACTION:y. P RECOND: uE FFECT: yWo I E. Wo I EI EYa. N?N,u I E. N?N,y. I EYa. I EYaN?N,u ? E. N?N,yN?N,yI EYa. N,i I E?a. N,iI EYaI yN,i I E?a. I y. N,i I E?aI EYa aI Wo I EYaN,i ? EE aN,i ? EYaN?N,u I EN,i I EThe trick is how exactly to do that. The Finish action in POP planning has as its preconditions the goal state. We can createIn this case there would be aFor example,By (11.1),Our axioms were of the formAction is possible. Rule. This tells us nothing about the case where the action isMore generally, if there are actionWith symbol-splitting, we don’t have to describe each specific flight, we need only sayMore generally, if there are action schemata of maximum arity, with u objects, and eachWith the notation used above, there areWith symbol-splitting, we wouldn’t gain anything for the oBut if we are willing to do some ratherPlanning and Acting in the Real. WorldEven if an action occupies a resource, that has no effect on theTherefore, we must extend the S TRIPS formalism. We could do this by treating a R ESOURCE: effect differently from other effects,We need to addI oGoI oGo. We addI oGo. I oGoThe important point is thatFor HireBuilder, the precondition is having the ability to pay,FillOutForm, and GetFormApproved. There is a causal link with the condition HaveFormFinally, the GetFormApproved step has the effect HavePermit. This is a valid decomposition. For HireBuilder, suppose we choose the three-step sequence InterviewBuilders, ChooseBuilder, and SignContract. This last step has the precondition AbleToPay and the effect HaveContractInHand. There are also causal links between the substeps, but they don’t affect theMostly these subplans are independent, but they must share the step of putting up a common post at the cornerNote that tasks are often decomposed specifically so as to minimize the amount of stepInstead they have “roughFor example, the decomposition of the LAToNYRoundTrip action can stipulate that the agent should go to New. York. In a simple STRIPS formulation where the start and goal states are the same, the emptyI I o a ?There remains the problem of preventing the STRIPS plan from including other stops on its itinerary; fixing this isThe “action” to be decomposed isUnfortunately, thereP RECOND: e. E FFECT: oAo-?I E?aI E aI E aAo-?That could be done with an action forE FFECT: e ?For the second case, where more than one instanceIt assumes that the outcomesIt is possible that a solutionHence, the space of possible actions sequences required toUnfortunately, moving affects these new literals.E FFECT: whenGoAlso, each forward move is typically short, followedThis is explained in terms of a disjunctiveThus, we have a strong cyclicThis is an unconditional plan, if taken literally, involving an infinite loop. If the preo?o uRinse and repeat ifThis is also a conditional step, although it is not specified here what problems are tested.There are three cases:WithEYaEYaE FFECT:when ?IaA new goal is simply addedBecause the data structures built byThere is no specific time bound that is guaranteed, and in general no such bound is possibleThere are two possible worlds, one where y holds and one where holds. In the first,UncertaintyIt is not enough to say thatFrom the definition of conditional probability, andA probability assignment to a set of propositions is consistentWe call theWe then have the following equations:EJaE5aUncertaintyEla. Therefore,IfFigure 13.3 loses if and ? are both true, the agent believes this to be impossible so the betThe number for which the agent is indifferent between the twoA set of degrees of belief specified for some collection of events will either be coherent,De Finetti showed that coherent beliefs satisfy the axioms of probabilityAxiom 1:If an agentThen the agent is indifferent between a payoffAxiom 3: Given two mutually exclusive, exhaustive events,I y I. I y IThis requires thatBy axioms 2 and 3, eachBy axiom 3, sinceThere are 13What is the probability ofOr you could assign a projectThe rest is easy, involving aP tCatch is true. First compute. P tWEJaUncertaintyRoughly speaking, if 10,000 people take the test, we expect 1 to actually have the disease, andMore precisely,A false positive reading remains much more likely. Here is an alternative exercise along the same lines: A doctor says that an infant whoIsabella predominantly turned her head to theIf it is 80 accurate? The reasoning is the same, and the answer is 50 right-handed if the test is 90 accurate, 69 right-handed if the test is 80 accurate.The second part of the exercise follows from by a similar derivation, or by noticing thatE aOf those, only 2 pick the fake coin, andCase 1: A fair coin might turn up headsThe probability of drawing the fake coin isChapter 14 describes generalpurpose, systematic algorithms that make heavy use of normalization. We could guess thatShowing that a given set of informationIntuitively, the information in (iii) is insufficient becauseMathematically, suppose u hasSimilarly, theThe information on the reliability ofWe need to know the probability that the taxi was blue, given that it looked blue. Thus we cannot decide the probability without some information about the prior probabilityGiven that 9 out of 10 taxis are green, and assuming the taxi in question is drawnIt is a variant of the “Monty Hall” problem, namedSeveral distinguished professors of statistics have very publically got the wrong answer. Certainly, allEGaOn the otherEGaUncertainty. Thus the key thing that is missed by the naive approach is that the guard has a choice of whomOne can produce variants of the question that reinforce the intuitions behind the correctPrisoner A finds a printout with the last page torn off. It givesWhat is the probability that A willWhat is the probability now. Clearly, in the first case it is quiteIt is this second case thatP eWe haveSimilarly, PFor example, the word pairIn general,Since thereTo calculate the probabilities of pits inWe can count the total number of 3-pit assignmentsI ? aThe second leaves 1 pit in the remaining 10 squares, so corresponds to 10 completeThe third also corresponds to 10 complete assignments.In 13.7(b), there are two partial assignments withYoWithYoYoProbabilistic ReasoningFormally speaking,Informally speaking, one neverInstead, one asks what variables are direct causes. It is usuallyN N,uThe following values indicate the general order ofBattery. Radio. IcyWeatherIgnitionMoves. CarFigureN N,uN N,u?EThat entry will be not quite 1 (because there is always someWe can relate noisy-AND to noisy-OR using de Morgan’s rule:In the noisy-OR case, we haveA?? y?AEJaEJaIt is exactly this kind of correlation that makes it difficult forThe wording of the question is a little tricky becauseProbabilistic ReasoningLiIt illustrates the magnitude of the achievement involved in creating complete and correctI o. I o. Abbreviating yI oThere are several ways to go about doing this. The “opportunistic” way is to noticeWe then use Bayes’ Rule again on the last term:Now we use the chain rule formula (Equation 15.1 on page 439) to rewrite the jointEGaE5aLettingEkaIf it is in focus, then we will assume there is aThe rest ofThen the table is as follows:Notice that each column has to add up to 1. Reasonable values forProbabilistic ReasoningOne approach uses the fact thatEither way, the solutions areI distribuI. AaI out-of-focus telescopeAaOne can also add in complexLet’s begin by looking at the multivariate Gaussian. From page 982 in Appendix A weThen, if we multiply out the exponent, we obtainI E5aWeI i IWoSimilar calculations yieldThe covariance matrixThe base case forThe inductive step asks us to show that if anyBy the product rule we haveI ELaExtending the argument of part (a), this is in turn a multivariate GaussianThe simplest way to do this is to take a linear combination ofStudents may need some help with the last part if they are to do it properly.The enumeration algorithm has two extra multiplications.For example,To eliminate anyThen, because theA Bayesian network corresponding to a SAT problem.The generalEach clauseA single sentence node has all the clauses as parents and a CPT that implementsHence, we have reduced. It is clear that. SAT to Bayes net inference. Since SAT is NP-complete, we have shown that Bayes netHence, we can count the numberOuThe we want is in the list with indexThen the expected runtimeEJaThe variance of the runtime can be reduced by further subdividing any part of the rangeWe can compute the table once and for all for the standard GaussianIf we had aInstead, one typically defines discrete ranges, returningProbably the simplest is to work directly from theNow, all terms in the product in the denominator that do not contain can be movedThis just leaves us with the terms that do mention, i.e., those in whichThen we haveStrictly speaking, the transition matrix is only well-defined for the variant of MCMC inEntries on the diagonal correspond to self-loops. Such transitions can occur byFor example,For example,Entries where both variables change cannot occur. For example,ForBoolean variables, the matrix is of sizeWN,y?oWith any prior onBC.Outcome. CA.Outcome. A Bayesian network corresponding to the soccer model of Exercise 14.12.