Artificial Intelligence Foundations Of Computational Agents 1st Edition David L. Poole, Alan K. Mackworth - Solutions

Discover comprehensive solutions to "Artificial Intelligence: Foundations of Computational Agents" by David L. Poole and Alan K. Mackworth. Access our online answers key and solutions manual to explore solved problems and chapter solutions. Enhance your understanding with our step-by-step answers and questions and answers guide. Our instructor manual and test bank offer in-depth insights, making it an invaluable resource for textbook enthusiasts. Download the solutions PDF for free and navigate through intricate concepts effortlessly. Unlock the potential of this textbook with our expertly crafted solutions and free download options, designed to support your academic journey.

3. Find out about two applications of AI (not classes of applications, but specific programs). For each application, write, at most, one typed page describing it. You should try to cover the following questions.(a) What does the application actually do (e.g., control a spacecraft, diagnose a
2. Give as many disciplines as you can whose aim is to study intelligent behavior of some sort. For each discipline, find out what aspect of behavior is investigated and what tools are used to study it. Be as liberal as you can regarding what defines intelligent behavior.
1. For each of the following, give five reasons why:(a) A dog is more intelligent than a worm.(b) A human is more intelligent than a dog.(c) An organization is more intelligent than an individual human.Based on these, give a definition of what “more intelligent” may mean.
• Inductive logic programming can be used to learn relational models, even when the values of features are meaningless names.
• Event calculus allows for continuous and discrete time and axiomatizes what follows from the occurrence of events.
• Situation calculus represents time in terms of the action of an agent, using the init constant and the do function.
• Many of the representations in earlier chapters can be made relational.
• Relational representations are used when an agent requires models to be given or learned before it knows what individuals it will encounter.
Exercise 14.8 Represent the electrical domain of previous chapters in ICL, so that it will run in AILog. The representation should include the probabilistic dependencies of Example 6.11 (page 238) and the relations of Example 12.11 (page 502).
Exercise 14.7 For the representation of addition in Example 14.20 (page 617), it was assumed that observed z-values would all be digits. Change the representation so that the observed values can be digits, a blank, or other. Give appropriate probabilities.
Exercise 14.6 Give some concrete specialization operators that can be used for top-down inductive logic programming. They should be defined so that making progress can be evaluated myopically. Explain under what circumstances the operators will make progress.
Exercise 14.5 Suppose that, in event calculus, there are two actions, Open and Close, and a relation opened that is initially, at time 0, false. Action Open makes opened true, and action Close makes opened false. Suppose that action Open occurs at time 5, and action Close occurs at time 10.(a)
Exercise 14.4 In this question, you will investigate using event calculus for the robot delivery domain.(a) Represent the move action in the event calculus.(b) Represent each of the sequences of actions in Example 14.10 (page 604) in the event calculus.(c) Show that event calculus can derive the
Exercise 14.3 AILog performs depth-bounded search. You will notice that the processing time for the previous questions was slow, and we required a depth bound that was close to the actual depth bound to make it work in a reasonable amount of time.In this question, estimate how long an iterative
Exercise 14.2 In this question, you will add a more complicated paint action than in previous exercise.Suppose the object paint can(Color) denotes a can of paint of color Color.Add the action paint(Obj, Color) that results in the object changing its color to Color. (Unlike the previous question,
Exercise 14.1 Add to the situation calculus example (also available from the course web page) the ability to paint an object. In particular, add the predicate color(Obj, Col, Sit)that is true if object Obj has color Col in situation Sit.The parcel starts off blue. Thus, we have an
Exercise 13.16 In this question, you will write a meta-interpreter for parametrized logic programs. These are logic programs that can use constants in arithmetic expressions. The values for the constants are given as part of the input to the meta-interpreter.Assume that an environment is a list of
Exercise 13.15 Build an iterative deepening abductive reasoning system to find minimal consistent sets of assumables to imply a goal. This can be based on the depth-bounded meta-interpreter of Figure 13.12 (page 587), and the delaying meta-interpreter of Figure 13.16 (page 590) to collect
Exercise 13.14 Write a meta-interpreter for definite clauses that does iterative deepening search. Make sure that it only returns one answer for each proof and that the system says no whenever the depth-first searcher says no. This should be based on the depth-bounded meta-interpreter (page 587)
Exercise 13.13 Write a meta-interpreter that allows both how and why questions.In particular, it should allow the user to ask how questions about a goal that has been proved after a why question. Explain how such a program may be useful.
Exercise 13.12 Write a program that takes in a tree produced from the metainterpreter that builds proof trees as shown in Figure 13.13 (page 587) and lets someone traverse the tree using how questions.
Exercise 13.11 Extend the ask-the-user meta-interpreter to have questions that ask for instances. You should be able to interpret declarations that say that a predicate is functional and respond accordingly.
Exercise 13.10 Write a meta-interpreter that allows for asking the users yes-orno questions. Make sure it does not ask questions to which it already knows the answer.
Exercise 13.9 Write a meta-interpreter that can ask multiple sources for information.Suppose that each source is identified by a universal resource identifier (URI).Suppose you have the predicates• can answer(Q, URI) is true if the source given by URI can answer questions that unify with Q.•
Exercise 13.8 The program of Figure 13.16 (page 590) allows duplicate delayed goals.Write a version of dprove that returns minimal sets of delayed goals, in their simplest forms.
Exercise 13.7 Modify the depth-bound meta-interpreter of Figure 13.12(page 587) so that(a) the bound is on the total length of the proof, where the length is the total number of instances of base-level atoms that appear in the proof.(b) different base-level atoms can incur different costs on the
Exercise 13.6 For the following, explain how each is categorized by BFO:(a) your skin(b) the period at the end of the previous sentence(c) the excitement a child has before a vacation(d) the trip home from a vacation(e) a computer program(f) summer holidays(g) the ring of a telephone(h) the dust on
Exercise 13.5 A luxury hotel has multiple rooms to rent, each of which is comfortable and has a view. The hotel must also have more than one restaurant. There must be menu items for vegetarians and for meat eaters to eat in the restaurants.(a) Define a luxury hotel in OWL, based on this
Exercise 13.4 Suppose a “beach resort” is a resort that is near a beach that the resort guests can use. The beach has to be near the sea or a lake, where swimming is permitted. A resort must have places to sleep and places to eat.Write a definition of beach resort on OWL.
Exercise 13.3 Write an ontology for the objects that often appear on your desk that may be useful for a robot that is meant to tidy your desk. Think of the categories that (a) the robot can perceive and (b) should be distinguished for the task.
Exercise 13.2 Sam has proposed that any n-ary relation P(X1,X2,X3, . . . ,Xn) can be reexpressed as n − 1 binary relations, namely, P1(X1,X2).P2(X2,X3).P3(X3,X4)....Pn−1(Xn−1,Xn).Explain to Sam why this may not be such a good idea. What problems would arise if Sam tried to do this?
Exercise 13.1 The aim of this exercise is to explore multiple representations for the same domain.Tic-tac-toe is a game played on a 3 × 3 grid. Two players, X and O, alternately place their marks in an unoccupied position. X wins if, during its turn, it can place an X in an unoccupied position to
• A meta-interpreter can be used to build a light-weight implementation of a knowledge-based system that can be customized to fit the requirements of the representation language.
• The ability to explain reasoning and debug knowledge in terms of semantic content can improve the usability of a knowledge-based system.
• OWL ontologies are built from individuals, classes, and properties. A class is a set of real and potential individuals. A property links an individual to a data type or an individual.
• Ontologies are required for knowledge sharing.
• Individual-property-value triples form a flexible, universal representation for relations.
• The independent choice logic allows for the specification of probabilistic models before the individuals are known.
• Inductive logic programming can be used to learn relational models, even when the values of features are meaningless names.
• Event calculus allows for continuous and discrete time and axiomatizes what follows from the occurrence of events.
• Situation calculus represents time in terms of the action of an agent, using the init constant and the do function.
• Many of the representations in earlier chapters can be made relational.
• Relational representations are used when an agent requires models to be given or learned before it knows what individuals it will encounter.
Exercise 12.17 Construct a knowledge base and a dictionary [based on Figure 12.11 (page 531)] to answer geographical questions such as that given in Figure 1.2(page 8). For each query, either show how it can be answered or explain why it is difficult to answer given the tools presented in this
Exercise 12.16 In this question, you are to write a definite clause knowledge base for the design of custom video presentations.Assume that the video is annotated using the relation segment(SegId, Duration, Covers), where SegId is an identifier for the segment. (In a real application this will be
Exercise 12.15 The aim of this question is to get practice writing simple logic programs.(a) Write a relation remove(E, L, R) that is true if R is the resulting list of removing one instance of E from list L. The relation is false if E is not a member of L.(b) Give all of the answers to the
Exercise 12.14 Consider the following logic program:ap(emp, L, L).ap(c(H, T), L, c(H, R)) ←ap(T, L, R).adj(A, B, L) ←ap(F, c(A, c(B, E)), L).(a) Give a top-down derivation (including all substitutions) for one answer to the query ask adj(b, Y, c(a, c(b, c(b, c(a, emp))))).(b) Are there any
Exercise 12.13 Consider the following logic program:rd(cons(H, cons(H, T)), T).rd(cons(H, T), cons(H, R)) ←rd(T, R).Give a top-down derivation, showing all substitutions for the query ask rd(cons(a, cons(cons(a,X), cons(B, cons(c, Z)))),W).What is the answer corresponding to this derivation?Is
Exercise 12.12 Consider the following logic program:f (empty,X,X).f (cons(X, Y),W, Z) ←f (Y,W, cons(X, Z)).Give each top-down derivation, showing substitutions (as in Example 12.31) for the query ask f (cons(a, cons(b, cons(c, empty))), L, empty).What are all of the answers?
Exercise 12.11 List all of the ground atomic logical consequences of the following knowledge base:q(Y) ← s(Y, Z) ∧ r(Z).p(X) ← q(f (X)).s(f (a), b).s(f (b), b).s(c, b).r(b).
Exercise 12.10 For each of the following pairs of atoms, either give a most general unifier or explain why one does not exist:(a) p(X, Y,a, b,W)p(E,c, F,G, F)(b) p(X, Y, Y)p(E, E, F)(c) p(Y,a, b, Y)p(c, F,G, F)(d) ap(F0, c(b, c(B0, L0)), c(a, c(b, c(b, c(a, emp)))))ap(c(H1, T1), L1, c(H1, R1))
Exercise 12.9 Give a most general unifier of the following pairs of expressions:(a) p(f (X), g(g(b))) and p(Z, g(Y))(b) g(f (X), r(X), t) and g(W, r(Q),Q)(c) bar(val(X, bb), Z) and bar(P, P)
Exercise 12.8 What is the result of the following application of substitutions:(a) f (A,X, Y,X, Y){A/X, Z/b, Y/c}.(b) yes(F, L) ← append(F, c(L, nil), c(l, c(i, c(s, c(t, nil))))){F/c(l,X1), Y1/c(L, nil),A1/l, Z1/c(i, c(s, c(t, nil)))}.(c) append(c(A1,X1), Y1, c(A1, Z1)) ← append(X1, Y1,
Exercise 12.7 Consider the following knowledge base:has access(X, library) ← student(X).has access(X, library) ← faculty(X).has access(X, library) ← has access(Y, library) ∧ parent(Y,X).has access(X, office) ← has
Exercise 12.6 In a manner similar to Example 12.21 (page 511), show derivations of the following queries:(a) ask two doors east(r107, R).(b) ask next door(R, r107).(c) ask west(R, r107).(d) ask west(r107, R).Give all answers for each query.
Exercise 12.5 In Example 12.21, we always selected the leftmost conjunct to resolve on. Is there a selection rule (a selection of which conjunct in the query to resolve against) that would have resulted in only one choice for this example? Give a general rule that – for this example, at least –
Exercise 12.4 In Example 12.21 (page 511), the algorithm fortuitously chose imm west(r109, r111) as the clause to resolve against. What would have happened if another clause had been chosen? Show the sequence of resolutions that arise,and either show a different answer or give a generalized answer
Exercise 12.3 Consider the following knowledge base:r(a).r(e).p(c).q(b).s(a, b).s(d, b).s(e, d).p(X) ← q(X) ∧ r(X).q(X) ← s(X, Y) ∧ q(Y).Show the set of ground atomic consequences derivable from this knowledge base.Assume that a bottom-up proof procedure is used and that at each iteration
Exercise 12.2 Consider the language that contains the constant symbolsa, b, and c; the predicate symbols p and q; and no function symbols. We have the following knowledge bases built from this language:KB1 = { p(a) }.KB2 = { p(X) ← q(X) }.KB3 = { p(X) ← q(X), p(a), q(b) }.Now consider possible
Exercise 12.1 Consider a domain with two individuals (✂and☎), two predicate symbols (p and q), and three constants (a,b, and c). The knowledge base KB is defined by p(X) ← q(X).q(a).(a) Give one interpretation that is a model of KB.(b) Give one interpretation that is not a model of KB.(c) How
• Clark’s completion can be used to define the semantics of negation as failure under the complete knowledge assumption.
• Equality between terms means that the terms denote the same individual.
• It is possible to use definite clauses to represent natural language grammars.
• Function symbols are used to denote a possibly infinite set of individuals described in terms of other individuals. Function symbols can be used to build data structures.
• Substitutions are used to make instances of atoms and rules. Unification makes atoms identical for use in proofs.
• Datalog is a logical language with constants, universally quantified variables, relations, and rules.
• In domains characterized by individuals and relations, constants denoting individuals and predicate symbols denoting relations can be reasoned with to determine what is true in the domain.
Exercise 11.12 In Example 11.14 (page 483), some of the features are perfectly correlated (e.g., F6 and F7). Does having such correlated features affect what functions can be represented? Does it help or hurt the speed at which learning occurs?
Exercise 11.11 In SARSA with linear function approximators, if you use linear regression to minimize r + γQw(s, a) − Qw(s, a), you get a different result than we have here. Explain what you get and why what is described in the text may be preferable (or not).
Exercise 11.10 Suppose your friend presented you with the following example where SARSA(λ) seems to give unintuitive results. There are two states, A and B. There is a reward of 10 coming into state A and no other rewards or penalties.There are two actions: left and right. These actions only make
Exercise 11.9 Consider four different ways to derive the value of αk from k in Qlearning(note that for Q-learning with varying αk, there must be a different count k for each state–action pair).i) Let αk = 1/k.ii) Let αk = 10/(9 + k).iii) Let αk = 0.1.iv) Let αk = 0.1 for the first 10,000
Exercise 11.8 Compare the different parameter settings for the game of Example 11.8 (page 464). In particular compare the following situations:(a) α varies, and the Q-values are initialized to 0.0.(b) α varies, and the Q-values are initialized to 5.0.(c) α is fixed to 0.1, and the Q-values are
Exercise 11.7 For the plot of the total reward as a function of time as in Figure 11.12 (page 474), the minimum and zero crossing are only meaningful statistics when balancing positive and negative rewards is reasonable behavior. Suggest what should replace these statistics when zero is not an
Exercise 11.6 Explain how Q-learning fits in with the agent architecture of Section 2.2.1 (page 46). Suppose that the Q-learning agent has discount factor γ, a step size of α, and is carrying out an -greedy exploration strategy.(a) What are the components of the belief state of the Q-learning
Exercise 11.5 Explain what happens in reinforcement learning if the agent always chooses the action that maximizes the Q-value. Suggest two ways to force the agent to explore.
Exercise 11.4 Suppose a Q-learning agent, with fixed α and discount γ, was in state 34, did action 7, received reward 3, and ended up in state 65. What value(s)get updated? Give an expression for the new value. (Be as specific as possible.)
Exercise 11.3 Give an algorithm for EM for unsupervised learning [Figure 11.4(page 457)] that does not store an A array, but rather recomputes the appropriate value for the M step. Each iteration should only involve one sweep through the data set. [Hint: For each tuple in the data set, update all
Exercise 11.2 Suppose the k-means algorithm is run for an increasing sequence of values for k, and that it is run for a number of times for each k to find the assignment with a global minimum error. Is it possible that a number of values of k exist for which the error plateaus and then has a large
Exercise 11.1 Consider the unsupervised data of Figure 11.1 (page 454).(a) How many different stable assignments of examples to classes does the kmeans algorithm find when k = 2? [Hint: Try running the algorithm on the data with a number of different starting points, but also think about what
• In reinforcement learning, an agent should trade off exploiting its knowledge and exploring to improve its knowledge.
• A Markov decision process is an appropriate formalism for reinforcement learning. A common method is to learn an estimate of the value of doing each action in a state, as represented by the Q(S,A) function.
• Missing values in examples are often not missing at random. Why they are missing is often important to determine.
• The probabilities and the structure of belief networks can be learned from complete data. The probabilities can be derived from counts. The structure can be learned by searching for the best model given the data.
• EM is an iterative method to learn the parameters of models with hidden variables (including the case in which the classification is hidden).
• By introducing payments, it is possible to design a mechanism that is dominant-strategy truthful and economically efficient.
• Agents can learn to coordinate by playing the same game repeatedly, but it is difficult to learn a randomized strategy.
• A Nash equilibrium is a strategy profile for each agent such that no agent can increase its utility by unilaterally deviating from the strategy profile.
• In partially observable domains, sometimes it is optimal to act stochastically.
• Perfect information games can be solved by backing up values in game trees or searching the game tree using minimax with α-β pruning.
• A multiagent decision network models probabilistic dependency and information availability.
• The extensive form of a game models agents’ actions and information through time in terms of game trees.
• The strategic form of a game specifies the expected outcome given controllers for each agent.
• A multiagent system consists of multiple agents who can act autonomously and have their own utility over outcomes. The outcomes depend on the actions of all agents. Agents can compete, cooperate, coordinate, communicate, and negotiate.
• A dynamic decision network allows for the representation of an MDP in terms of features.
• A fully observable MDP can be solved with value iteration or policy iteration.
• An MDP can represent an infinite stage or indefinite stage sequential decision problem in terms of states.
• A decision network can represent a finite stage partially observable sequential decision problem in terms or features.
• Utility is a measure of preference that combines with probability.
• Different planning algorithms can be used to convert a planning problem into a search problem.
• An action is a function from a state to a state. A number of representations exploit structure in representation of states. In particular, the feature-based representation of actions represents what must be true in the previous state for a feature to have a value in the next state. The STRIPS

Showing 2700 - 2800 of 4756

First
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Last

Artificial Intelligence Foundations Of Computational Agents 1st Edition David L. Poole, Alan K. Mackworth - Solutions

Step by Step Answers