Questions and Answers of Artificial Intelligence Structures

Implement the unify algorithm of Section 2.3.2 in the computer language of your choice.Data from section 2.3.2To apply inference rules such as modus ponens, an inference system must be able to
Take the logic-based financial advisor of Section 2.4, put the predicates describing the problem into clause form, and use resolution refutations to answer queries such as whether a particular
We discussed the card game Poker briefly in Section 13.3. Given that the present (probabilistic) state of a player is either good bet, questionable bet, or bad bet, work out a POMDP to represent this
How would you change the MDP representation of Section 13.3 to a POMDP? Take the simple robot problem and its Markov transition matrix created in Section 13.3.3 and change it into a POMDP. Think of
The area of agent-based research was introduced in Section 12.3. We recommend further reading on any of the projects mentioned, but especially the Brooks, Nilsson and Benson, or Crutchfield and
Show how the add and delete lists can be used to replace the frame axioms in the generation of STATE 2 from STATE 1 in Section 8.4.Data from state 2Data from state 1 ontable(a). ontable(c).
Create the remaining frame axioms necessary for the four operator's pickup, putdown, stack, and unstack described in rules 4 through 7 of Section 8.4.Data described from rule 4 through 7. 4. (VX)
Take Exercise 1 above. Create in English or pseudocode 15 if-then rules (other than those prescribed in Section 8.2) to describe relations within this domain. Create a graph to represent the
Use of the stochastic approach for discovering patterns in a relational database is an important area of current research, sometime referred to as data mining (see Section 10.3). How might this work
Assume that managers are listed in the employee_salary relation with other employees in the example of Section 15.5.2. Extend the example so that it will handle queries such as “Find any employee
How might the stochastic approaches of Section 15.4 be combined with the techniques for database analysis found in Section 15.5.2.Data from section 15.4Data from section 15.5.2 In Section 15.1 we
Extend the database front end example of Section 15.5.2 so that it will answer questions of the form “How much does Don Morrison earn?” You will need to extend the grammar, the representation
Describe how the Markov models of Section 15.4 might be combined with the more symbolic approach to understanding language of Section 15.1−15.3. Describe how the stochastic models could be combined
Define concepts and relations in conceptual graphs needed to represent the meaning of the grammar of Exercise 9. Define the procedures for building a semantic representation from the parse tree.Data
Define an ATN parser for the dog's world grammar with adjectives (Exercise 7) and prepositional phrases (Exercise 8).Data from exercise 7 Extend the dogs world grammar to include adjectives in noun
Add the following context-free grammar rules to the dog's world grammar of Section 15.2.1. Map the resulting grammar into transition networks. sentence ↔ noun_phrase verb_phrase
Use resolution for queries in the farmer, wolf, goat, and cabbage problem of Section 15.3.Data from section 15.3 To this point we have offered representations and algorithms that support symbol based
How would you do data-driven reasoning with resolution? Use this to address the search space of Exercise 1. What problems might arise in a large problem space?Data from exercise 1Take the logic-based
In Chapter 5 we presented a simplified form of the knight’s tour. Take the path 3 rule, put it in clause form, and use resolution to answer queries such as path 3 (3,6). Next, use the recursive
Use resolution to answer the query in Example 3.3.4.Data from Example 3.3.4This example is taken from the predicate calculus and represents a goal-driven graph search where the goal to be proved true
Section 11.5.4, used the linear associator algorithm to make two vector pair associations. Select three (new) vector pair associations and solve the same task. Test whether your linear associator is
Select a different input pattern than that we used in Section 11.5.2. Use the unsupervised Hebbian learning algorithm to recognize that pattern.Data from section 11.5.2 Recall that in unsupervised
Analyze Samuel’s checker playing program from a reinforcement learning perspective. Sutton and Barto (1998, Section 11.2) offer suggestions in this analysis.
What happens if the temporal difference algorithm of Problem 13 plays tic-tac-toe against itself?Data from problem 13Consider the tic-tac-toe example of Section 10.7.2. Implement the temporal
Consider the tic-tac-toe example of Section 10.7.2. Implement the temporal difference learning algorithm in the language of your choice. If you designed the algorithm to take into account problem
Write a program that implements the fuzzy controller of Section 9.2.2.Data from section 9.2.2 There are two assumptions that are essential for the use of formal set theory. The first is with respect
Use the schema axioms presented in McCarthy (1980, Section 4) to create the circumscription results presented in Section 9.1.3.Data from section 9.1.3In the previous sections, we extended logic by
Expand the propositional calculus representation scheme introduced by Williams and Nayak for describing state transitions for their propulsion system.
Read Williams and Nayak (1996, 1996a) for a more complete discussion of their modelbased planning system.
Show two more incompatible (precondition) subgoals in the blocks world operators of Figure 8.19.Figure 8.19 b a c a U A b d C U 4 8.8 b d d a a b m C A 0.80 d C Figure 8.19 Portion of the state
Read and comment on the survey paper Improving Human Decision Making through Case-Based Decision Aiding by Janet Kolodner (1991).
Pick another area of interest for designing an expert system. Answer Exercises 1–3 for this application.Data from exercise 1In Section 8.2 we introduced a set of rules for diagnosing automobile
Consider the graph of Exercise 2. Do you recommend data-driven or goal-driven search? breadth-first or depth-first search? In what ways could heuristics assist the search? Justify your answers to
Pick one of the application areas for agent architectures summarized in Section 7.4. Choose a research or application paper in that area. Design an organization of agents that could address the
Examples of analogical reasoning were presented in Section 7.3.2. Describe an appropriate representation and search strategy that would allow for identification of the best answer for this type
Translate the financial advisor knowledge base, Section 2.4, into conceptual graph form.Data from section 2.4 As a final example of the use of predicate calculus to represent and reason about problem
Repeat problem 9.b to produce a goal-driven solution.Data from problem 9.bb. Generate the state space and stages of working memory for the data-driven solution to the example in Chapter 3.
Perform a left-to-right alpha-beta prune on the tree of Exercise 13. Perform a right-to-left prune on the same tree. Discuss why a different pruning occurs.Data from exercise 13Put the following
Prove that more informed heuristics develop the same or less of the search space. formalize the argument presented in Section 4.3.3.Data from section 4.3.3 The final issue of this subsection compares
Prove that the set of states expanded by algorithm A* is a subset of those examined by breadth-first search.
Add grammar rules to Example 3.3.6 that allow complex sentences such as, sentence ↔ sentence AND sentence.Data from Example 3.3.6Our final example is not from the predicate calculus but consists of
Add rules for (multiple) prepositional phrases to the English grammar of Example 3.3.6.Data from Example 3.3.6Our final example is not from the predicate calculus but consists of a set of rewrite
Add rules defining adjectives and adverbs to the grammar of Example 3.3.6.Several interesting problems from earlier editions of our book were left out of the 4th edition of Chapter 3. They are
Trace a data-driven execution of the financial advisor of Example 3.3.5 for the case of an individual with four dependents, $18,000 in the bank, and a steady income of $25,000 per year. Based on a
Choose and justify a choice of breadth - or depth-first search for examples of Exercise 6.Data from exercise 6Implement a backtrack algorithm in a programming language of your choice.
Given the example of the Viterbi algorithm processing the probabilistic finite state machine of Figure 13.7 and 13.8: Figure 13.7Figure 13.8 a. Why is new seen as a better interpretation than knee
Expand the ATN grammar of Section 15.2.4 to include who and what questions.
Extend the context-sensitive grammar of Section 15.2.3 to test for semantic agreement between the subject and verb. Specifically, men should not bite dogs, although dogs can either like or bite men.
Parse each of these sentences using the context-sensitive grammar of Section 15.2.3.The men like the dog.The dog bites the man.
Extend the dogs world grammar so it will include the illegal sentences in Exercise 3.Data from Exercise 3Parse each of these sentences using the “dogs world” grammar of Section 15.2.1. Which of
Discuss the representational structures and knowledge necessary to understand the following sentences.The brown dog ate the bone.Attach the large wheel to the axle with the hex nut.Mary watered the
Use factoring and resolution to produce a refutation for the following clauses: p(X) ∨ p(f(Y)) and ¬ p(W) ∨ ¬ p(f(Z)). Try to produce a refutation without factoring.
Create the and/or graph for the following problem. Why is it impossible to conclude the goal:r(Z) ∨ s(Z)?Fact: p(X) ∨ q(X).Rules: p(a) → r(a) and q(b) → s(b).
Prove the linear input form strategy is not refutation complete.
Create the and/or graph for the following data-driven predicate calculus deduction.Fact: ¬ d(f) ∨ [b(f) ∧ c(f)].Rules: ¬ d(X)→¬a(X) and b(Y) → e(Y) and g(W) ← c(W).Prove: ¬ a(Z) ∨
Put the following predicate calculus expression in clause form: ∀ (X)(p(X) → {∀ (Y)[p(Y) → p (f (X, Y))] ∧ ¬ ∀ (Y) [q (X, Y) → p(Y)]})
Take the happy student problem of Figure 14.5 and apply three of the refutation strategies of Section 14.2.4 to its solution.Figure 14.5Data from section 14.2.4
Pick a “canonical set” of six family relations. Write demodulators to reduce alternative forms of relations to the set. For example, your “mother’s brother” is “uncle.”
Write a demodulator for sum that would cause clauses of the form equal (ans, sum (5, sum (6, minus (6)))) to be reduced to equal (ans, sum (5, 0)). Write a further demodulator to reduce this last
Work out two examples for hyperresolution where the nucleus has at least four literals.
Use resolution to solve the following puzzle from Wos et al. (1984). Four people: Roberta, Thelma, Steve, and Pete hold eight different jobs, each person with exactly two jobs. The jobs are, chef,
How might you use resolution to implement a “production system” search?
Use resolution to prove Wirth’s statement in Exercise 12, Chapter 2.Data from Exercise 12The following story is from N. Wirth’s (1976) Algorithms + data structures = programs. I married a widow
Work out the complexity cost for finding an optimal policy for the POMDP problem exhaustively.
Jack has a car dealership and is looking for a way to maximize his profits. Every week, Jack orders a stock of cars, at the cost of d dollars per car. These cars get delivered instantly. The new cars
Program the MDP supported robot of Section 13.3.3 in the language of your choice. Experiment with different values of a and b that can optimize the reward. There are several interesting possible
Hand run the robot described in the Markov decision process example of Section 13.3.3. Use the same reward mechanism and select probabilistic values for a and b for the decision processing.a. Run
Given the hidden Markov model and Viterbi algorithm of Section 9.3.6, perform a full trace, including setting up the appropriate back pointers, that shows how the observation #, n, iy, t, # would be
Create a figure to represent the hierarchical hidden Markov model of Section 13.1.2. What type problem situation might your HHMM be appropriate for? Discuss the issue of fitting the structures of
Create a hidden Markov model to represent the scoring sequence of an American football game where touchdowns are 6 points which can be followed by either a 0, 1, or 2 extra point attempt. Of course,
For further insights into evolution and complexity, read and discuss Darwin’s Dangerous Idea (Dennett 1995) or Full House: The Spread of Excellence from Plato to Darwin (Gould 1996).
Discuss the role of inductive bias in the representations, search strategies, and operators used in the models of learning presented in Chapter 12. Is this issue resolvable? That is, does the genetic
Write an a-life program that implements the functionality of Figures 12.10−12.13.Figure 12.10 Figure 12.11Figure 12.12Figure 12.13
Read the early discussion of the Game of Life in Gardner’s column of Scientific American (1970, 1971). Discuss other a-life structures, similar to the glider, of Section 12.3.1.
Discuss the constraints on using genetic programming techniques to solve problems. For example, what components of a solution cannot be evolved within the genetic programming paradigm?
Write a program to solve Kepler’s Third Law of Motion Problem, described with a preliminary representation offered in Section 12.2.2.
How does the Bucket Brigade Algorithm (Holland 1986) relate to the back propagation algorithm (Section 14.3)?
Read Holland’s Schema Theorem (Mitchell 1996, Koza 1992). How does Holland’s schema theory describe the evolution of the GA solution space? What does it have to say about problems not encoded as
Build a genetic algorithm to search for a solution for the traveling salesperson problem.
Consider the traveling salesperson problem of Section 12.1.3. Discuss the problem of selecting an appropriate representation for this problem. Design other appropriate genetic operators and fitness
Build a genetic algorithm in the language to solve the CNF-satisfaction problem.
Consider the CNF-satisfaction problem of Section 12.1.3. How does the role of the number of disjuncts in the CNF expression bias the solution space? Consider other possible representations and
Discuss the problem of designing representations for genetic operators to search for solutions in different domains? What is the role of inductive bias here?
The genetic algorithm is intended to support the search for genetic diversity along with the survival of important skills (represented by genetic patterns) for a problem domain. Describe how
Write a Hopfield net to solve the traveling salesperson problem for ten cities.
Describe the differences between the BAM memory and the linear associator. What is crosstalk and how can it be prevented?
Consider the bidirectional associative memory (BAM) of Section 11.6.3. Change the association pairs given in our example and create the weight matrix for the associations. Select new vectors and test
Use a backpropagation net to recognize the ten (hand drawn) digits. One approach would be to build a 4 x 6 array of points. When a digit is drawn on this grid it will cover some elements, giving them
Write a counter propagation net to solve the exclusive-or problem. Compare your results with those of backpropagation net of Section 11.3.3. Use your counter propagation net to discriminate between
Write a Kohonen net in LISP or C++ and use it to classify the data of Table 11.3. Compare your results with those of Sections 11.2.2 and 11.4.2.Table 11.3
Build a backpropagation network in LISP or C++ and use it to solve the exclusive-or problem of Section 11.3.3. Solve the exclusive-or problem with a different backpropagation architecture, perhaps
Build a perceptron net in LISP and run it on the classification example of Section 11.2.2.a. Generate another data set similar to that of Table 11.3 and run your classifier on it. Table
Make a McCulloch−Pitts neuron that can calculate the logic function implies, ⇒.
Another problem type excellent for reinforcement learning is the so-called grid world. We present a simple 4 x 4 grid world in Figure 10.26. The two greyed corners are the desired terminal states for
Can you analyze the inverted pendulum problem, Figure 8.8, presented in Section 8.2.2 from a reinforcement learning perspective? Build some simple reward measures and use the temporal difference
Develop an explanation-based learning algorithm in the language of your choice. If you use Prolog, consider the algorithms developed in the auxiliary material.
Develop a domain theory for explanation-based learning in some problem area of your choice. Trace the behavior of an explanation-based learner in applying this theory to several training instances.
From Quinlan (1993) obtain the C4.5 decision tree algorithm and test it on a data set.There are complete programs and data sets for C4.5 available from this reference.
Other problems of ID3 are bad or missing data. Data is bad if one set of attributes has two different outcomes. Data is missing if part of the attribute is not present, perhaps because it was too
Discuss problems that can arise from using continuous attributes in data, such as a monetary cost, dollars and cents, or the height, a real number, of an entity. Suggest some method for addressing

Showing 1 - 100 of 247